BenchLeader
Menu
Choose a modelLeaderboardsModelsBenchmarksProvidersReleases10

Every benchmark.
One leaderboard.

74 benchmarks from 22 independent publishers, reconciled into one set of standings every morning. BenchLeader runs no evaluations of its own: it lines up what Artificial Analysis, LMArena, Epoch AI, LiveBench, SWE-bench and the rest have already measured, so every model is judged on all of the evidence at once, at the reasoning effort it was actually run at.

Compare modelsBest valueFastestProviders
models ranked
431
configurations ranked
756
sources live
23/23

As of 5 Oct 2026, BenchLeader ranks GPT-6 Astra first at 70.2, ahead of Claude Opus 5.5 (70.1) and Claude Fable 5.1 (69.9).

  • Claude Opus 5.5 (high) moved from #7 to #2 in the BenchLeader Index.
  • Claude Opus 5.5 (thinking) moved from #2 to #3 in the BenchLeader Index.
  • Claude Opus 5.5 (max) moved from #3 to #4 in the BenchLeader Index.
  • Claude Fable 5.1 (xhigh) moved from #4 to #6 in the BenchLeader Index.
  • Read the daily digest
Standings5 Oct 2026
  1. 1GPT-6 AstramaxOpenAI · $20.00/M70.2▼ 0.1
  2. 2Claude Opus 5.5highAnthropic · $8.00/M70.1▲ 0.4
  3. 3Claude Fable 5.1xhighAnthropic · $20.00/M69.9▼ 0.1
  4. 4NewGemini 4 ArgonhighGoogle · $4.00/M69.7▲ 0.2
  5. 5NewGPT-6.1 SolmaxOpenAI · $4.00/M69.5▲ 0.2
  6. 6Claude Fable 5thinkingAnthropic · $20.00/M69.2▼ 0.5
  7. 7Claude Opus 5highAnthropic · $10.00/M68.4▼ 0.2
  8. 8Muse Spark 1.3maxMeta · $2.00/M67.9▲ 0.1
  9. 9GPT-5.6 SolmaxOpenAI · $8.00/M67.5▼ 0.1
  10. 10NewClaude Sonnet 5.5xhighAnthropic · $4.00/M67.0▼ 1.8
Purple: leader. Arrows: change in index since the last time it moved. Full standings

NewThis week

Release watch

And 4 more on Release watch.

The frontier board

The 10 models that matter this week, read three ways. Click a lane to re-order.

Model
  1. 1GPT-6 Astramax70.254$0.80
  2. 2Claude Opus 5.5high70.174$0.32
  3. 3Claude Fable 5.1xhigh69.966$0.80
  4. 4NewGemini 4 Argonhigh69.7–$0.16
  5. 5NewGPT-6.1 Solmax69.563$0.16
  6. 6Muse Spark 1.3max67.9152$0.09
  7. 7MiMo-V2.6-Pro64.846$0.03
  8. 8DeepSeek V4 Promax62.9107$0.05
  9. 9GLM 5.3 Flash62.454$0.01
  10. 10Qwen3.8-Flash-Next59.954$0.01
Bars are relative to the best in the lane; purple is the lane leader.Cost is one agentic coding session, 60k tokens in and 4k out.Full quality boardFull speed boardFull cost board

Quality against…

Showing 20 models: the frontier no other model beats on both axes, the top eight by index, the best model in each price band, and anything new in the last seven days. Up and to the left is better: more quality for less money. Hover for details, click to open a model.

GPT-6 AstramaxClaude Opus 5.5highClaude Fable 5.1xhighGemini 4 ArgonhighGPT-6.1 SolmaxClaude Fable 5thinkingClaude Opus 5highMuse Spark 1.3maxClaude Sonnet 5.5xhighMiMo-V2.6-ProDeepSeek V4 PromaxGLM 5.3 FlashQwen3.8-Flash-NextGPT-6 LunamaxMiMo-V2.6-FlashStep 3.5 FlashLing 3.0 FlashQwen3.5-9BthinkingQwen3.5-4BthinkingGranite 4.2 3B
506070$0.050$0.100$0.200$0.500$1.00$2.00$5.00$10.00$20.00Blended price, $ per 1M tokens (log scale)BenchLeader IndexGPT-6 Astra (max)Claude Opus 5.5 (high)Claude Fable 5.1 (xhigh)NEW Gemini 4 Argon (high)NEW GPT-6.1 Sol (max)Claude Fable 5 (thinking)Claude Opus 5 (high)Muse Spark 1.3 (max)NEW Claude Sonnet 5.5 (xhigh)MiMo-V2.6-ProDeepSeek V4 Pro (max)GLM 5.3 FlashQwen3.8-Flash-NextGPT-6 Luna (max)MiMo-V2.6-FlashStep 3.5 FlashLing 3.0 FlashQwen3.5-9B (thinking)Qwen3.5-4B (thinking)Granite 4.2 3B
  • Anthropic5
  • Alibaba3
  • OpenAI3
  • Xiaomi2
  • Ant Group1
  • DeepSeek1
  • Google1
  • IBM1
  • Meta1
  • StepFun1
  • Zhipu AI1
Dashed line: the frontier. Purple-ringed points are not beaten on both axes by any other model shown. Dashed ring: new in the last seven days.

The BenchLeader Index

One row per model, showing its best-scoring reasoning effort; switch to all configurations to compare efforts. Click any column to sort. Methodology

431 of 431
#
1GPT-6 AstramaxOpenAI70.2±3$20.0054342 s1.1M2026-09-022026-10-02
2Claude Opus 5.5highAnthropic70.1±5$8.007437 s1M2026-09-222026-10-02
3Claude Fable 5.1xhighAnthropic69.9±6$20.0066112 s1M2026-09-012026-09-01
4NewGemini 4 ArgonhighGoogle69.7±4$4.00––1M2026-09-302026-10-02
5NewGPT-6.1 SolmaxOpenAI69.5±3$4.0063291 s1.1M2026-09-292026-10-02
6Claude Fable 5thinkingAnthropic69.2±7$20.006697 s1M2026-06-092026-06-09
7Claude Opus 5highAnthropic68.4±4$10.005420 s1M2026-07-242026-10-02
8Muse Spark 1.3maxMeta67.9±4$2.0015265 s1.0M2026-09-022026-10-02
9GPT-5.6 SolmaxOpenAI67.5±4$8.0080113 s1.1M2026-07-092026-09-29
10NewClaude Sonnet 5.5xhighAnthropic67.0±6$4.0010532 s1M2026-09-282026-10-02
11GPT-6 SolmaxOpenAI66.3±3$4.0093141 s1.1M2026-09-222026-10-02
12GPT-5.5xhighOpenAI65.9±4$11.258552 s1.1M2026-04-232026-10-02
13Kimi K3maxMoonshot AIopen ↗65.6±3$6.003464 s1.0M2026-07-162026-10-02
14GPT-5.5 ProxhighOpenAI65.6±9$67.5093.20 s1.1M2026-04-232026-04-23
15Qwen3.8 MaxmaxAlibaba65.0±3$3.003954 s1M2026-08-032026-10-02
16MiMo-V2.6-ProXiaomiopen ↗64.8±5$0.5444648 s1.0M2026-09-212026-10-02
17Gemini 3.8 FlashmediumGoogle64.7±5$1.50832.47 s1.0M2026-09-022026-09-02
18Muse Spark(Apr 2026)Meta64.4±3–––262k2026-04-082026-10-02
19Grok 4.6mediumSpaceXAI64.3±5$3.007624 s500k2026-08-112026-08-12
20GPT-5.4xhighOpenAI64.2±3$5.6389193 s1.1M2026-03-052026-09-29
21GLM 5.3maxZhipu AIopen ↗64.1±3$2.157132 s1M2026-08-182026-10-02
22GPT-5.3-CodexxhighOpenAI64.0±5$4.819569 s400k2026-02-052026-09-29
23Claude Opus 4.7Anthropic63.8±2$10.00481.61 s1M2026-04-162026-10-02
24GPT-5.6 TerramaxOpenAI63.7±5$4.50100177 s1.1M2026-07-092026-09-29
25Grok 4.7highSpaceXAI63.6±6$3.007833 s500k2026-09-212026-09-21
26Muse Spark 1.1Meta63.5±6$2.002203.55 s1.0M2026-07-092026-10-02
27Qwen3.8 2.4T A95BAlibabaopen ↗63.5±5$3.004053 s262k2026-08-12–
28Gemini 3.7 FlashmediumGoogle63.4±3$1.502835.62 s1.0M2026-08-132026-08-13
29GPT-5.4 ProOpenAI63.4±2$67.5046.09 s1.1M2026-03-042026-03-23 · stale
30Claude Opus 4.6Anthropic63.1±5$10.00381.54 s1M2025-12-172026-10-02
31Gemini 3.5 FlashmediumGoogle63.1±3$3.3821114 s1.0M2026-05-192026-10-02
32DeepSeek V4 PromaxDeepSeekopen ↗62.9±3$0.99010743 s1M2026-08-132026-09-29
33GLM 5.2maxZhipu AIopen ↗62.7±4$2.158427 s1M2026-06-132026-10-02
34Muse Spark 1.2xhighMeta62.7±4$2.0024222 s1.0M2026-08-052026-10-02
35Gemini 3.1 ProGoogle62.6±4$4.5011725 s1.0M2026-03-052026-10-02
36Qwen3.6 MaxmaxAlibaba62.5±3$2.927330 s262k2026-04-202026-10-02
37Claude Opus 4.8maxAnthropic62.4±5$10.006163 s1M2026-05-282026-06-25
38GLM 5.3 FlashZhipu AIopen ↗62.4±5$0.2385440 s1M2026-08-262026-10-02
39DeepSeek V4.1 FlashmaxDeepSeekopen ↗61.7±3$0.26220911 s1M2026-09-102026-10-02
40Step 5 PreviewStepFun61.6±5$1.438626 s1M2026-09-182026-10-02
41GPT-5.2xhighOpenAI60.9±3$4.8180120 s128k2025-12-112026-09-29
42Gemini 3 ProGoogle60.8±3––––2025-11-182026-10-02
43DeepSeek V4 Flash(Apr 2026)maxsuperseded ↗DeepSeekopen ↗60.7±5$0.26221610 s1M2026-04-242026-04-24
44Claude Sonnet 5highAnthropic60.6±2$4.006910 s1M2026-06-302026-10-02
45Gemini 3.6 FlashhighGoogle60.5±4$1.5018918 s1.0M2026-07-212026-10-02
46Claude Opus 4.5thinkingAnthropic60.2±3$10.004917 s200k2025-11-242026-09-29
47GPT-5 ProOpenAI60.2±6$41.25––400k2025-08-072025-11-13 · stale
48Deepseek v4 Pro HighhighDeepSeek60.1±5––––2026-04-242026-10-02
49Qwen3 7maxAlibaba59.9±4$3.7520314 s1M2026-05-192026-10-02
50Qwen3.8-Flash-NextAlibabaopen ↗59.9±6$0.2305440 s256k2026-08-262026-10-02