# Benchmark Heaven > The most detailed cost–capability analysis in AI. Every model. Every Benchmark. Actual Costs. Benchmark results for large language models, each with its source and date, and a modeled cost per task that accounts for provider prices, caching and the model's own token use. Home of JevBench, the benchmark for AI decision models, including Jev-compatible systems. ## Pages - [Home](https://benchmarkheaven.com/): recommended models by capability and cost - [Benchmarks](https://benchmarkheaven.com/benchmarks): every benchmark board with sources - [Compare models](https://benchmarkheaven.com/compare): side-by-side benchmark results and costs - [Providers](https://benchmarkheaven.com/providers): per-provider prices, caching and speed - [Sources and methodology](https://benchmarkheaven.com/about): where the data comes from and how cost is modeled ## JevBench (AI decision models, including Jev-compatible systems) - [JevBench leaderboard](https://benchmarkheaven.com/jev-models): open-weights Jev-compatible decision models by intelligence, calibration, speed and cost, with two unranked Jev reference rows: the Jev version that defines the Jev-class caps, and TypeSafe's current Jev version (ranked on the API board) - [JevBench API leaderboard](https://benchmarkheaven.com/jev-models/api): hosted decision APIs (Jev, wity, Sage, Fastino and more) ranked on the same scores - [ImageJevBench leaderboard](https://benchmarkheaven.com/image-jev-bench): image decision systems by intelligence, calibration, speed and cost, inside the same cost and latency budget as JevBench - [AudioJevBench](https://benchmarkheaven.com/audio-jev-bench): audio models that make typed voice-agent decisions (intent, sentiment, urgency, speaker verification, sound events) straight from audio - [Jev alternatives](https://benchmarkheaven.com/jev-models/alternatives) - [Jev-Alternativen im Vergleich (Deutsch)](https://benchmarkheaven.com/de/jev-models/alternativen) - [Is Jev open source? Open-source Jev-class models](https://benchmarkheaven.com/jev-models/open-source-jev) - [How to choose a Jev-class model](https://benchmarkheaven.com/jev-models/how-to-choose) - [Jev vs Imajev-4B](https://benchmarkheaven.com/jev-models/jev-vs-imajev) - [Jev vs Plumb-4B](https://benchmarkheaven.com/jev-models/jev-vs-plumb) - [Jev vs decider-4b v2](https://benchmarkheaven.com/jev-models/jev-vs-decider-4b-v2) - [Jev vs JevK5](https://benchmarkheaven.com/jev-models/jev-vs-jevk5) - [Jev vs Cygnet](https://benchmarkheaven.com/jev-models/jev-vs-cygnet) - [Jev vs Hopper](https://benchmarkheaven.com/jev-models/jev-vs-hopper) - [Jev vs Winnow-12B Q8](https://benchmarkheaven.com/jev-models/jev-vs-winnow-12b-q8) - [Jev vs reflex 4B](https://benchmarkheaven.com/jev-models/jev-vs-reflex-4b) - [Jev vs Laya](https://benchmarkheaven.com/jev-models/jev-vs-laya) - [Jev vs Xor 26B-A4B](https://benchmarkheaven.com/jev-models/jev-vs-xor-26b-a4b) - [Jev vs Surogate Rune 26B-A4B v3](https://benchmarkheaven.com/jev-models/jev-vs-surogate-rune-26b-a4b-v3-v16) - [Jev vs Torchcast Decision 12B](https://benchmarkheaven.com/jev-models/jev-vs-torchcast-decision-12b) - [Jev vs Jev-Omni](https://benchmarkheaven.com/jev-models/jev-vs-jev-omni) - [Jev vs Quyet-1.0-Large](https://benchmarkheaven.com/jev-models/jev-vs-quyet-1-0-large) - [Jev vs decisio v0.8.0 on gemma-4-31B-it](https://benchmarkheaven.com/jev-models/jev-vs-decisio-gemma-4-31b-v080) - [Jev vs Sage 1.3.0](https://benchmarkheaven.com/jev-models/jev-vs-sage-1.3.0) - [Jev vs deck-31B](https://benchmarkheaven.com/jev-models/jev-vs-deck31b) - [Jev vs René-1 31B FP8](https://benchmarkheaven.com/jev-models/jev-vs-rene-1-31b-fp8) - [Jev vs H2O-Lightning-4B v1.1](https://benchmarkheaven.com/jev-models/jev-vs-h2o-lightning-4b) - [Jev vs SPX-CD Flash](https://benchmarkheaven.com/jev-models/jev-vs-spx-cd-flash) - [Jev vs kettei-fat](https://benchmarkheaven.com/jev-models/jev-vs-kettei-fat) - [Jev vs Chon-1.0-12B](https://benchmarkheaven.com/jev-models/jev-vs-chon-1.0-12b) - [Jev vs H2O-Lightning-31B v1.0.7](https://benchmarkheaven.com/jev-models/jev-vs-h2o-lightning-31b) - [Jev vs Drex Max](https://benchmarkheaven.com/jev-models/jev-vs-drex-max) - [Jev vs basal-omni-1.0-ultra](https://benchmarkheaven.com/jev-models/jev-vs-basal-omni-1.0-ultra) - [Jev vs Jeff-1.0-Large](https://benchmarkheaven.com/jev-models/jev-vs-jeff_1_0_large) - [Jev vs SubQ Snap](https://benchmarkheaven.com/jev-models/jev-vs-subq-snap-preview) - Model pages: https://benchmarkheaven.com/jev-models/{system} (replace {system} with the public model slug) - [Full JevBench data and method](https://benchmarkheaven.com/llms-full.txt) - [Decision model benchmarks compared](https://benchmarkheaven.com/decision-model-benchmarks): JevBench, Hanno Labs DecisionBench, RewardBench, ForecastBench, tau-bench and the separate metamorphic coherence JevBench - [Methodology](https://benchmarkheaven.com/jev-models/methodology): Capability versus gated harmonic Composite, model versions and reproducibility limits - [Data card and own JSON/CSV measurements](https://benchmarkheaven.com/jev-models/data) - [Calibration and reliability](https://benchmarkheaven.com/jev-models/calibration-reliability) - [Evaluating Jev-compatible alternatives](https://benchmarkheaven.com/jev-models/evaluate-alternatives) ## Data and API - [Public API reference](https://github.com/fstandhartinger/model-market-comparison/blob/main/API.md) - Current JevBench release: v1.6.39 (always read the live value from the feed below, field latest_release) - [JevBench base-release manifest](https://benchmarkheaven.com/api/jevbench/manifest): the base release (v1.6.1) the live board is built on, with scoring definitions, weights and source hashes; for the current release use https://benchmarkheaven.com/api/jevbench/latest (field latest_release) - [Current JevBench leaderboard, JSON](https://benchmarkheaven.com/api/jevbench/latest): revision, weights, systems with keys, sources, axes, Capability eligibility/rank, Composite score/rank, listing flags and price basis - [Models, JSON](https://benchmarkheaven.com/api/models) ## For agents - This is JevBench by Benchmark Heaven at benchmarkheaven.com. JevBench at jevbench.github.io is a separate metamorphic coherence project. - Capability is the open-weights headline; Composite leads the API board. Equal axis weights feed a gated harmonic mean, not an arithmetic average. Historical scores keep their own method. - Start with the current JevBench JSON feed. Capability is the mean of Intelligence and Calibration; use capability.eligible and capability.rank for the official cost and median-latency caps. - Pick by an axis when that priority dominates. For custom weights, use the published axes and the documented gated harmonic Composite formula; custom scores are exploratory, not official ranks. Higher axis scores are better; lower actual price and latency are better. - MCP server (read-only, no auth): https://benchmarkheaven.com/api/mcp — Streamable HTTP, one JSON-RPC message per POST; tools list_models, get_model and compare_models over the same JevBench feed. Pages also register these tools through WebMCP (document.modelContext) where the browser supports it. - Compare two ranked model keys: https://benchmarkheaven.com/jev-models?compare=a,b#compare (replace a,b with feed keys). - [Submit a model](https://benchmarkheaven.com/submit): free queue or fast lane - See the Public API reference above for schema, units, pricing basis and missing-value rules. ## Optional - [Source code (MIT; third-party data keeps its own terms)](https://github.com/fstandhartinger/model-market-comparison)