SearchBook a call

BULLETIN / BENCHMARKS

Benchmarks, from across the internet.

14 models, 12 independent evaluations, 44 scores. Every number links to the page it was read from.

Scores were read from the named evaluations on 2026-09-30. Higher is better unless the column says otherwise. A blank cell means the evaluation had not published a score for that model, or the review only had a vendor-reported figure. Vendor-reported numbers are not on this page; each review discusses them separately.

ModelAA IndexArtificial AnalysisArena textArena (arena.ai)Vals IndexVals AIEpoch ECIEpoch AIFrontierMath T1-3Epoch AIGPQA DiamondEpoch AIARC-AGI-2ARC PrizeArena WebDevArena (arena.ai)Arena visionArena (arena.ai)Terminal-Bench 4.0Artificial AnalysisHLE (AA)Artificial AnalysisHLE (Scale)Scale LabsPrice inUSD/M tokensPrice outUSD/M tokensContexttokens
Claude Opus 5.5Anthropic · closed58max effort150966.97%·91.2%90.6%93.3%1820·59.6%61.4%·4201M tokens
GPT-6.1 SolOpenAI · closed52max effort···········2101,050,000 tokens
Gemini 3.8 FlashGoogle · closed41high effort149254.83%157.1value·95.4%89.2%····44.520.753.751,048,576 input tokens
Grok 4.7xAI · closed46high and xhigh effort both scored 46·54.95%····1636xhigh····26500,000 tokens
Muse Spark 1.3Meta · closed48max effort·58.16%max effort156.974.4%xhigh effort··16551290max effort···1.254.251M tokens
Mistral Medium 3.5Mistral AI · open weights14···········1.57.5256k tokens
DeepSeek-V4.1-FlashDeepSeek · open weights39max effort·51.32%·········0.301.201M tokens
Qwen3.8-MaxAlibaba Cloud (Qwen) · closed450902 snapshot·48.27%····16711301···261,000,000 tokens
Kimi K3Moonshot AI · open weights44148850.70%···60.4%1659····3.0015.001,048,576 tokens
Gemma 4 (E2B, E4B, 12B Unified, 26B A4B, 31B)Google · open weights1931B model··142.731B model··········128K tokens (E2B, E4B); 256K tokens (12B, 26B A4B, 31B)
gpt-oss-120bOpenAI · open weights12high effort·············131,072 tokens
Granite 4.2 (3B, 8B, 30B)IBM · open weights··············128K tokens native, extendable to 512K (8B model card)
Sarvam 30B and Sarvam 105BSarvam AI · open weights··············65,536 tokens (Sarvam 30B)
MiMo-V2.6-ProXiaomi · open weights46148055.20%···········1M tokens

Where the numbers come from

Prices and context windows come from each vendor's own documentation, as cited in the reviews. Scores change as evaluations add runs; the date at the top is the date every cell was read.