Benchmarks
More
OverviewBenchmaxxingCompareChartsCost vs CapabilityProviders per ModelProvider explorerGatewaysEU & SovereignJev-class modelsImage JevBenchAudioJevBenchSubmit a modelAbout

Jev vs baselabs mammoth-1: published benchmark comparison

Data: JevBench v1.8.0, published 2026-10-29 · method v1.7 (10 s rule), re-scored from the recorded per-decision times

Rank #1 on the open-weights board in v1.8.0.

Compare Jev-compatible models by separate measures and their measurement conditions. A historical measurement is not a current rank.

Jev and baselabs mammoth-1: published axes and category radars

Pick any two. Radars for the score axes, capability by subject topic and use cases, chance-corrected competence per request type on the open and sealed sets, and competence per tier on each set. Further out is better on every spoke; the link keeps the pair.

  • A: baselabs mammoth-1 — Open weights · LLM decoder · Score 75.4 (#1)
  • B: Jev 1.13.0REFERENCE · TypeSafe — Jev — reference (TypeSafe, closed) · Score 71.4 (ranked, not ranked)

The four score axes

Radar: the four score axes, two systemsThe four score axes, baselabs mammoth-1 vs Jev 1.13.0. Intelligence: 76.2 vs 63.6; Calibration: 94.3 vs 90.6; Speed: 84.4 vs 91.2; Cost: 57.2 vs 54.7.50100Intelligence: baselabs mammoth-1: 76.2; Jev 1.13.0: 63.6Intelligence76.2 · 63.6Calibration: baselabs mammoth-1: 94.3; Jev 1.13.0: 90.6Calibration94.3 · 90.6Speed: baselabs mammoth-1: 84.4; Jev 1.13.0: 91.2Speed84.4 · 91.2Cost: baselabs mammoth-1: 57.2; Jev 1.13.0: 54.7Cost57.2 · 54.7
0–100, the values in the table. An axis a system has no published value for is left as a gap (it counts as 0 in the composite).

Capability by subject topic

baselabs mammoth-1: No per-category values are published for this system.

Radar: capability by subject topic, two systemsCapability by subject topic, baselabs mammoth-1 vs Jev 1.13.0. Rules, policy & law: No per-category values are published for this system. vs 49.4; Coding & software: No per-category values are published for this system. vs 64.3; Math & numbers: No per-category values are published for this system. vs 10.4; Finance & commerce: No per-category values are published for this system. vs 40.9; Support & operations: No per-category values are published for this system. vs 48.8; Everyday language: No per-category values are published for this system. vs 77.5; Safety & security: No per-category values are published for this system. vs 52.2. Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.−1000100Rules, policy & law: baselabs mammoth-1: No measured result; Jev 1.13.0: 49.4 — Rules, policy & law: applying written rules: company policies, contracts, regulations, eligibility. 1616 items (151 open, 1465 sealed).Rules, policy& lawn/a · 49.4Coding & software: baselabs mammoth-1: No measured result; Jev 1.13.0: 64.3 — Coding & software: code, SQL, repositories, developer tools and IT systems. 382 items (62 open, 320 sealed).Coding &softwaren/a · 64.3Math & numbers: baselabs mammoth-1: No measured result; Jev 1.13.0: 10.4 — Math & numbers: a calculation decides the answer: arithmetic, word problems, probability, dates, units. 370 items (37 open, 333 sealed).Math &numbersn/a · 10.4Finance & commerce: baselabs mammoth-1: No measured result; Jev 1.13.0: 40.9 — Finance & commerce: money: payments, refunds, invoices, orders, expenses, insurance payouts. 223 items (12 open, 211 sealed).Finance &commercen/a · 40.9Support & operations: baselabs mammoth-1: No measured result; Jev 1.13.0: 48.8 — Support & operations: support tickets, incidents, logistics, scheduling desks and routing work to a team. 160 items (21 open, 139 sealed).Support &operationsn/a · 48.8Everyday language: baselabs mammoth-1: No measured result; Jev 1.13.0: 77.5 — Everyday language: short everyday messages: intents, assistant requests, reading a detail out of a text. 145 items (11 open, 134 sealed).Everydaylanguagen/a · 77.5Safety & security: baselabs mammoth-1: No measured result; Jev 1.13.0: 52.2 — Safety & security: untrusted or injected instructions, fraud, moderation, access and security triage. 109 items (6 open, 103 sealed).Safety &securityn/a · 52.2

Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.

What the decision is about — math and dates, coding, rules, money, support work, everyday messages, safety. Chance-corrected competence per category (0 = at or below chance; negative averages are reported as 0, 100 = perfect), baselabs mammoth-1: open+sealed; Jev 1.13.0: S+P+L1+L2+L3 items pooled; only categories with at least 30 items are spokes, smaller ones are listed below. Hover a category for its definition and item count.
What each category means · items per category
  • Rules, policy & law — applying written rules: company policies, contracts, regulations, eligibility. 1616 items (151 open / 1465 sealed)
  • Coding & software — code, SQL, repositories, developer tools and IT systems. 382 items (62 open / 320 sealed)
  • Math & numbers — a calculation decides the answer: arithmetic, word problems, probability, dates, units. 370 items (37 open / 333 sealed)
  • Finance & commerce — money: payments, refunds, invoices, orders, expenses, insurance payouts. 223 items (12 open / 211 sealed)
  • Support & operations — support tickets, incidents, logistics, scheduling desks and routing work to a team. 160 items (21 open / 139 sealed)
  • Everyday language — short everyday messages: intents, assistant requests, reading a detail out of a text. 145 items (11 open / 134 sealed)
  • Safety & security — untrusted or injected instructions, fraud, moderation, access and security triage. 109 items (6 open / 103 sealed)

Use cases (TypeSafe categories)

baselabs mammoth-1: No per-category values are published for this system.

Radar: use cases (typesafe categories), two systemsUse cases (TypeSafe categories), baselabs mammoth-1 vs Jev 1.13.0. Model routing: No per-category values are published for this system. vs 73.0; Legal & compliance: No per-category values are published for this system. vs 68.6; Customer support: No per-category values are published for this system. vs 49.6; Other: No per-category values are published for this system. vs 40.1; E-commerce: No per-category values are published for this system. vs 40.5; Insurance claims: No per-category values are published for this system. vs 46.1; Risk assessment: No per-category values are published for this system. vs 32.8; Financial crime: No per-category values are published for this system. vs 35.8; Feature extraction: No per-category values are published for this system. vs 28.3; Lead generation: No per-category values are published for this system. vs 49.2; Recruiting: No per-category values are published for this system. vs 0.0; Knowledge graphs: No per-category values are published for this system. vs 28.8; LLM guardrails: No per-category values are published for this system. vs 68.0; Moderation: No per-category values are published for this system. vs 51.6; Code linting: No per-category values are published for this system. vs 29.1; Search & retrieval: No per-category values are published for this system. vs 80.9; Science: No per-category values are published for this system. vs 21.2; Advertising: No per-category values are published for this system. vs 16.4; Gaming: No per-category values are published for this system. vs 5.1; Demand forecasting: No per-category values are published for this system. vs 0.0. Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.−1000100Model routing: baselabs mammoth-1: No measured result; Jev 1.13.0: 73.0 — Model routing: deciding which LLM or model tier should handle a prompt; classifying a request's intent, domain, difficulty or risk to pick a model. 524 items (86 open, 438 sealed).Modelroutingn/a · 73.0Legal & compliance: baselabs mammoth-1: No measured result; Jev 1.13.0: 68.6 — Legal and compliance: contracts, policies, regulations, eligibility rules, checking documents against written requirements. 515 items (74 open, 441 sealed).Legal &compliancen/a · 68.6Customer support: baselabs mammoth-1: No measured result; Jev 1.13.0: 49.6 — Customer support: classifying tickets and customer intents, urgency, refunds, routing cases to a team, checking support replies. 240 items (37 open, 203 sealed).Customersupportn/a · 49.6Other: baselabs mammoth-1: No measured result; Jev 1.13.0: 40.1 — Other: none of the above: general reasoning, arithmetic or date puzzles, everyday assistant requests without a business workflow. 174 items (24 open, 150 sealed).Othern/a · 40.1E-commerce: baselabs mammoth-1: No measured result; Jev 1.13.0: 40.5 — E-commerce marketplaces: product listings, product attributes, orders and deliveries, seller catalog policy. 119 items (9 open, 110 sealed).E-commercen/a · 40.5Insurance claims: baselabs mammoth-1: No measured result; Jev 1.13.0: 46.1 — Insurance claims: first-notice-of-loss reports, claim complexity, missing information, claim triage. 119 items (9 open, 110 sealed).Insuranceclaimsn/a · 46.1Risk assessment: baselabs mammoth-1: No measured result; Jev 1.13.0: 32.8 — Risk assessment: estimating probabilities and severity of risks, incidents, vendor or underwriting risk. 118 items (8 open, 110 sealed).Riskassessmentn/a · 32.8Financial crime: baselabs mammoth-1: No measured result; Jev 1.13.0: 35.8 — Financial crime: suspicious transactions, KYC, fraud alerts, entity matching for investigations. 101 items (9 open, 92 sealed).Financialcrimen/a · 35.8Feature extraction: baselabs mammoth-1: No measured result; Jev 1.13.0: 28.3 — Feature extraction for predictive modeling: turning natural-language data into probabilistic features or estimates for a downstream prediction. 99 items (5 open, 94 sealed).Featureextractionn/a · 28.3Lead generation: baselabs mammoth-1: No measured result; Jev 1.13.0: 49.2 — Lead generation: matching companies or inbound messages to an ideal customer profile, buyer intent, prioritising and routing sales leads. 95 items (4 open, 91 sealed).Leadgenerationn/a · 49.2Recruiting: baselabs mammoth-1: No measured result; Jev 1.13.0: 0.0 — Recruiting: resumes, applications, interview feedback, matching candidates to roles. 94 items (1 open, 93 sealed).Recruitingn/a · 0.0Knowledge graphs: baselabs mammoth-1: No measured result; Jev 1.13.0: 28.8 — Graphs and knowledge graphs: entity types, relationships between records, contradictions between facts, multi-step lookups across linked facts. 94 items (3 open, 91 sealed).Knowledgegraphsn/a · 28.8LLM guardrails: baselabs mammoth-1: No measured result; Jev 1.13.0: 68.0 — LLM guardrails: checking LLM inputs, outputs and tool calls: prompt injection, jailbreaks, policy violations, tool-call errors, answer quality. 93 items (5 open, 88 sealed).LLMguardrailsn/a · 68.0Moderation: baselabs mammoth-1: No measured result; Jev 1.13.0: 51.6 — Moderation and trust and safety: user content moderation: toxicity, harassment, spam, unsafe advice, personal data exposure. 93 items (7 open, 86 sealed).Moderationn/a · 51.6Code linting: baselabs mammoth-1: No measured result; Jev 1.13.0: 29.1 — Semantic code linting: checking code or writing against conventions and guidelines, as in CI review of code. 90 items (3 open, 87 sealed).Code lintingn/a · 29.1Search & retrieval: baselabs mammoth-1: No measured result; Jev 1.13.0: 80.9 — Search and retrieval: scoring query-to-candidate relevance, reranking results, selecting useful context for RAG. 90 items (5 open, 85 sealed).Search &retrievaln/a · 80.9Science: baselabs mammoth-1: No measured result; Jev 1.13.0: 21.2 — Scientific discovery: screening papers, labelling research passages or survey answers, checking that citations support claims, methodology checks. 90 items (2 open, 88 sealed).Sciencen/a · 21.2Advertising: baselabs mammoth-1: No measured result; Jev 1.13.0: 16.4 — Advertising: ad creatives, campaign copy, brand safety, prohibited claims in ads. 88 items (3 open, 85 sealed).Advertisingn/a · 16.4Gaming: baselabs mammoth-1: No measured result; Jev 1.13.0: 5.1 — Gaming: player reports, in-game chat, game support. 86 items (3 open, 83 sealed).Gamingn/a · 5.1Demand forecasting: baselabs mammoth-1: No measured result; Jev 1.13.0: 0.0 — Demand forecasting: purchase intent, product interest, demand and supply signals for forecasting. 83 items (3 open, 80 sealed).Demandforecastingn/a · 0.0

Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.

The real-world application area of each decision, after the TypeSafe use-case map; an item can count in two. Chance-corrected competence per category (0 = at or below chance; negative averages are reported as 0, 100 = perfect), baselabs mammoth-1: open+sealed; Jev 1.13.0: S+P+L1+L2+L3 items pooled; only categories with at least 30 items are spokes, smaller ones are listed below. Hover a category for its definition and item count.A point on the bold 0 ring printed 0.0 is a measured value, not a gap: these published cells clip below-chance results to 0, so 0 means at or below chance. Their score markers cannot fall inside that ring.
What each category means · items per category
  • Model routing — deciding which LLM or model tier should handle a prompt; classifying a request's intent, domain, difficulty or risk to pick a model. 524 items (86 open / 438 sealed)
  • Legal and compliance — contracts, policies, regulations, eligibility rules, checking documents against written requirements. 515 items (74 open / 441 sealed)
  • Customer support — classifying tickets and customer intents, urgency, refunds, routing cases to a team, checking support replies. 240 items (37 open / 203 sealed)
  • Other — none of the above: general reasoning, arithmetic or date puzzles, everyday assistant requests without a business workflow. 174 items (24 open / 150 sealed)
  • E-commerce marketplaces — product listings, product attributes, orders and deliveries, seller catalog policy. 119 items (9 open / 110 sealed)
  • Insurance claims — first-notice-of-loss reports, claim complexity, missing information, claim triage. 119 items (9 open / 110 sealed)
  • Risk assessment — estimating probabilities and severity of risks, incidents, vendor or underwriting risk. 118 items (8 open / 110 sealed)
  • Financial crime — suspicious transactions, KYC, fraud alerts, entity matching for investigations. 101 items (9 open / 92 sealed)
  • Feature extraction for predictive modeling — turning natural-language data into probabilistic features or estimates for a downstream prediction. 99 items (5 open / 94 sealed)
  • Lead generation — matching companies or inbound messages to an ideal customer profile, buyer intent, prioritising and routing sales leads. 95 items (4 open / 91 sealed)
  • Recruiting — resumes, applications, interview feedback, matching candidates to roles. 94 items (1 open / 93 sealed)
  • Graphs and knowledge graphs — entity types, relationships between records, contradictions between facts, multi-step lookups across linked facts. 94 items (3 open / 91 sealed)
  • LLM guardrails — checking LLM inputs, outputs and tool calls: prompt injection, jailbreaks, policy violations, tool-call errors, answer quality. 93 items (5 open / 88 sealed)
  • Moderation and trust and safety — user content moderation: toxicity, harassment, spam, unsafe advice, personal data exposure. 93 items (7 open / 86 sealed)
  • Semantic code linting — checking code or writing against conventions and guidelines, as in CI review of code. 90 items (3 open / 87 sealed)
  • Search and retrieval — scoring query-to-candidate relevance, reranking results, selecting useful context for RAG. 90 items (5 open / 85 sealed)
  • Scientific discovery — screening papers, labelling research passages or survey answers, checking that citations support claims, methodology checks. 90 items (2 open / 88 sealed)
  • Advertising — ad creatives, campaign copy, brand safety, prohibited claims in ads. 88 items (3 open / 85 sealed)
  • Gaming — player reports, in-game chat, game support. 86 items (3 open / 83 sealed)
  • Demand forecasting — purchase intent, product interest, demand and supply signals for forecasting. 83 items (3 open / 80 sealed)

Competence per request type, open / sealed

Radar: competence per request type, open / sealed, two systemsCompetence per request type, open / sealed, baselabs mammoth-1 vs Jev 1.13.0. Choice · open: 88.9 vs 79.0; Choice · sealed: 82.9 vs 75.8; Noul · open: 72.2 vs 54.5; Noul · sealed: 72.3 vs 46.1; Score · open: 73.0 vs 63.3; Score · sealed: 67.9 vs 63.0. Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.−1000100Choice · open: baselabs mammoth-1: 88.9; Jev 1.13.0: 79.0Choice · open88.9 · 79.0Choice · sealed: baselabs mammoth-1: 82.9; Jev 1.13.0: 75.8Choice ·sealed82.9 · 75.8Noul · open: baselabs mammoth-1: 72.2; Jev 1.13.0: 54.5Noul · open72.2 · 54.5Noul · sealed: baselabs mammoth-1: 72.3; Jev 1.13.0: 46.1Noul · sealed72.3 · 46.1Score · open: baselabs mammoth-1: 73.0; Jev 1.13.0: 63.3Score · open73.0 · 63.3Score · sealed: baselabs mammoth-1: 67.9; Jev 1.13.0: 63.0Score ·sealed67.9 · 63.0

Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.

Chance-corrected competence for Choice, Noul and Score on the 300 open and 1200 sealed decisions.

Competence per tier — open set

Radar: competence per tier — open set, two systemsCompetence per tier — open set, baselabs mammoth-1 vs Jev 1.13.0. Easy: 92.9 vs 93.9; Standard: 72.2 vs 63.2; Judge: 73.8 vs 67.7; Hard: 85.6 vs 65.6. Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.−1000100Easy: baselabs mammoth-1: 92.9; Jev 1.13.0: 93.9Easy92.9 · 93.9Standard: baselabs mammoth-1: 72.2; Jev 1.13.0: 63.2Standard72.2 · 63.2Judge: baselabs mammoth-1: 73.8; Jev 1.13.0: 67.7Judge73.8 · 67.7Hard: baselabs mammoth-1: 85.6; Jev 1.13.0: 65.6Hard85.6 · 65.6

Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.

Per-tier competence, the three request types pooled by their published decision counts.

Competence per tier — sealed set

Radar: competence per tier — sealed set, two systemsCompetence per tier — sealed set, baselabs mammoth-1 vs Jev 1.13.0. Easy: 87.4 vs 78.4; Standard: 78.6 vs 68.7; Judge: 77.3 vs 66.9; Hard: 71.8 vs 58.9. Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.−1000100Easy: baselabs mammoth-1: 87.4; Jev 1.13.0: 78.4Easy87.4 · 78.4Standard: baselabs mammoth-1: 78.6; Jev 1.13.0: 68.7Standard78.6 · 68.7Judge: baselabs mammoth-1: 77.3; Jev 1.13.0: 66.9Judge77.3 · 66.9Hard: baselabs mammoth-1: 71.8; Jev 1.13.0: 58.9Hard71.8 · 58.9

Linear signed scale: −100 at the centre, 0 on the bold middle ring, 100 at the rim.

Per-tier competence on the sealed decisions, types pooled the same way; item text stays private.
How the categories were made

O1S chance-corrected competence: equal mean across the request types present, clipped to 0–100. Zero means the equal-weight mean across present request types was at or below baseline; negative means are clipped. It does not mean every type scored zero. Choice uses a random-option baseline; Noul uses a fixed 50% accuracy baseline (abstentions count as wrong); Score uses the midpoint-guess error (random-guess fallback when every gold is at the midpoint). Compare models within a category; row pools and supported types can differ.

Categories use ruled subject-topic labels and authoring use-case precedence across the answered pool union, with stable item identities counted once.

  • Official O1S; recorded failures count where supported; no missing response is imputed; stable identity dedupe parent first.
  • S-based category cells pool recorded supported responses from S+P and the answered L1/L2/L3 supplements. API overlay category cells keep their separately recorded A4/A5+P union. Headline and per-type/tier scores are unchanged.
All values as a table
SpokeA: baselabs mammoth-1B: Jev 1.13.0
The four score axes
Intelligence76.263.6
Calibration94.390.6
Speed84.491.2
Cost57.254.7
Capability by subject topic
Rules, policy & lawNo per-category values are published for this system.49.4
Coding & softwareNo per-category values are published for this system.64.3
Math & numbersNo per-category values are published for this system.10.4
Finance & commerceNo per-category values are published for this system.40.9
Support & operationsNo per-category values are published for this system.48.8
Everyday languageNo per-category values are published for this system.77.5
Safety & securityNo per-category values are published for this system.52.2
Use cases (TypeSafe categories)
Model routingNo per-category values are published for this system.73.0
Legal & complianceNo per-category values are published for this system.68.6
Customer supportNo per-category values are published for this system.49.6
OtherNo per-category values are published for this system.40.1
E-commerceNo per-category values are published for this system.40.5
Insurance claimsNo per-category values are published for this system.46.1
Risk assessmentNo per-category values are published for this system.32.8
Financial crimeNo per-category values are published for this system.35.8
Feature extractionNo per-category values are published for this system.28.3
Lead generationNo per-category values are published for this system.49.2
RecruitingNo per-category values are published for this system.0.0
Knowledge graphsNo per-category values are published for this system.28.8
LLM guardrailsNo per-category values are published for this system.68.0
ModerationNo per-category values are published for this system.51.6
Code lintingNo per-category values are published for this system.29.1
Search & retrievalNo per-category values are published for this system.80.9
ScienceNo per-category values are published for this system.21.2
AdvertisingNo per-category values are published for this system.16.4
GamingNo per-category values are published for this system.5.1
Demand forecastingNo per-category values are published for this system.0.0
Competence per request type, open / sealed
Choice · open88.979.0
Choice · sealed82.975.8
Noul · open72.254.5
Noul · sealed72.346.1
Score · open73.063.3
Score · sealed67.963.0
Competence per tier — open set
Easy92.993.9
Standard72.263.2
Judge73.867.7
Hard85.665.6
Competence per tier — sealed set
Easy87.478.4
Standard78.668.7
Judge77.366.9
Hard71.858.9

Published values

MeasureJev 1.13.0 (TypeSafe AI)baselabs mammoth-1 (27B)
Capability Score77.185.2
Rank (open-weights board)Reference, not ranked#1
Composite (secondary)71.475.4
intelligence63.676.2
calibration90.694.3
speed91.284.4
cost54.757.2
USD per 1,000 decisions$0.032 per 1,000 decisions$0.027 per 1,000 decisions
p50 latency (seconds)0.20.188 measured on-GPU, no network · 0.53 est. incl. network (2x + 150 ms)
Methodv1.7 · 10 s rulev1.7 · 10 s rule
p95 (measured) · decisions over 10 sp95 0.311 s (measured) · 0 of 1,500 over 10 sp95 0.270 s (measured) · 0 of 1,500 over 10 s
Before the 10 s rule (v1.6)v1.6.43: A 71.49 · #18—
Last measured2026-10-052026-10-22

* Self-hosted rows: the first value is the p50 we measured on our own hardware over loopback (on-GPU, or on-CPU where marked, no network). It is not directly comparable with API latency, which we time end to end from our measuring host, network included. “est.” is the est. incl. network (2x + 150 ms) that the ranking and the latency cap use for every self-hosted row. Model authors can ask us to time a hosted endpoint instead (on request).

Jev 1.13.0 (TypeSafe AI)

proprietary API

Published source
Measured on
the operator's hosted API
Cost basis (not published)
TypeSafe list tariff USD 0.042 per 1M input tokens, output free; measured on the v1.6.1 common cost basis (all items except those outside the Jev reference input range) from this run's token usage
Latency adjustment
none (API)

baselabs mammoth-1 (27B)

MIT (model card metadata since 21 Oct 2026; no LICENSE file; base Qwen/Qwen3.5-27B is Apache-2.0), public, not gated

Published source
Measured on
Evaluator-owned rented Verda GPU VM, one NVIDIA RTX PRO 6000 Blackwell Server Edition 96 GB (driver 580.178.04), FIN-03 (Finland, EU), pay-as-you-go non-spot, instance type 1RTXPRO6000.30V at USD 2.549/h. The launcher and engine run in a docker container with network none (cut proven from inside before any item hash or model load: loopback only, no default route, DNS and TCP fail), read-only root, tmpfs scratch, read-only model and input, empty credential variables, HF offline flags, one request at a time, timed over loopback on the VM. Peak CUDA memory allocated after the scored run 59,375 MiB of 97,887 MiB. Measured 22 Oct 2026 (row 09:57-10:06Z, scored run 401 s) on the frozen v1.6-regular-20261012 self-hosted input (sha256 dfd36720..., checked on the node and inside the container). This row's input was shredded from the node and its absence verified, and the VM was removed afterwards.
Cost basis (estimated)
GPU-time estimate from v1.8.0: one monthly-rented RTX PRO 6000 96 GB, EUR 999/month / 730 hours, ECB USD 1.1133/EUR dated 28 Oct 2026, USD 1.5235/hour; average load 70%. Architecture: dense qwen, 27B active parameters; throughput 11840.1727884 tokens/s (direct measurement, 29 Oct 2026, stock vLLM 0.32.0 EFF, fresh 200 public decisions; load_mode=base-arch). The row's own scored cost basis: 777,323 input + 1,485 output tokens over 1,485 decisions = 524.44983165 tokens/decision. GPU-time cost USD 0.0267785975171361/1,000; old list-price cost USD 0.103632717171717/1,000; applied min(old list-price cost, GPU-time cost) = USD 0.0267785975171361/1,000. The GPU-time estimate is lower. Estimated, not a bill. Intelligence, Calibration, measured latency and the common cost-item membership are unchanged. Prior prices remain in the archived v1.7 releases.
Latency adjustment
x2 + 0.15 s (assumption, not measured)

Using these models through an API

Decision Models offers a Jev-compatible API; its comparison page lists hosted options and prices, checked on the date shown there.

Jev-compatible APIs compared (Decision Models)

Decision Models is run by the same company as Benchmark Heaven. This link is the same on every comparison page and never affects scores or ranks.

Frequently asked questions

What does JevBench show for Jev and baselabs mammoth-1?
Jev 1.13.0 has Capability 77.1 and is the unranked reference (the open-weights board ranks 217 open-weights systems; hosted API offerings are ranked on the API leaderboard). baselabs mammoth-1 (27B) has Capability 85.2. Rank #1 on the open-weights board in v1.8.0. Historical scores use their original method and are not ranked against the current release.
Is there a time limit per decision?
Yes, from JevBench v1.7.0 (method v1.7): a decision that takes longer than 10 seconds (measured time) counts as a wrong answer and as zero calibration for that item. Each row shows its p95 and how many decisions were over the limit; the v1.6 values stay visible.
How does Capability differ from Composite?
Capability is the mean of Intelligence and Calibration; headline ranks require both official cost and median-latency caps. Composite additionally scores Speed and Cost and is secondary.
Can I compare the prices as actual bills?
Jev 1.13.0 (TypeSafe AI): not published, $0.032 per 1,000 decisions. baselabs mammoth-1 (27B): estimated, $0.027 per 1,000 decisions. Estimates and self-reported vendor prices are not measured charges.
Is baselabs mammoth-1 open source?
The published licence note for baselabs mammoth-1 (27B) is MIT (model card metadata since 21 Oct 2026; no LICENSE file; base Qwen/Qwen3.5-27B is Apache-2.0), public, not gated. Jev 1.13.0 is listed as proprietary API. Check the linked sources for terms.