Frontier Log

Benchmark · InactiveNo longer reported by the major labs since Jul 2026

MCP Atlas

Created by Scale AI, 19 Sep 2025 · Higher is better · Website

020406080100Oct 2025Jan 2026Apr 2026Jul 2026Oct 202685.8%Opus 571.67%Ripple-65B Preview75.3%GPT-5.567.2%GPT-5.462.3%Opus 4.547.6%GPT-5 mini020406080100Jan 2026Jul 202685.8%Opus 571.67%Ripple-65B Preview79.7%Opus 4.767.2%GPT-5.462.3%Opus 4.5

Results

ModelScoreSource
Claude Opus 524 Jul 2026 · Anthropic85.8%Claude Opus 5 System CardChecked A person checked this number against its source on 28 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.
Claude Fable 59 Jun 2026 · Anthropic83.3%Claude Fable 5 and Claude Mythos 5 System CardChecked A person checked this number against its source on 28 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.
LongCat-2.51 Nov 2026 · Meituan83.1%LongCat-2.5: Native Multimodality for Agentic TasksChecked A person checked this number against its source on 2 Nov 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.max reasoning effort; 256K contextOur capture
Claude Opus 4.828 May 2026 · Anthropic82.2%Claude Opus 4.8 System CardCorroborated The same number is printed in 1 other lab document: 1Run by Scale AI · 100-tool-call budget; adaptive thinking, max effort
Claude Opus 4.716 Apr 2026 · Anthropic79.7%Claude Opus 4.7 System CardChecked A person checked this number against its source on 28 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.Run by Scale AI · extended config of 256 turns and 100 tools; high effort
Claude Opus 4.716 Apr 2026 · Anthropic79.5%Claude Opus 4.7 System CardCorroborated The same number is printed in 1 other lab document: 1Run by Scale AI · extended config of 256 turns and 100 tools; max effort; another report, not charted
Claude Opus 4.716 Apr 2026 · Anthropic79.1%Claude Opus 4.8 System CardChecked A person checked this number against its source on 28 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.Run by Scale AI · 100-tool-call budget and updated judge; reported later; another report, not charted
GLM-5.216 Jun 2026 · Z.ai77.0%GLM-5.2: Built for Long-Horizon TasksChecked A person checked this number against its source on 19 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.maximum thinking effort; the post's chart, where its table prints 76.8Source image
Inkling15 Jul 2026 · Thinking Machines76.0%Inkling: Our Open-Weights ModelCorroborated The same number is printed in 2 other lab documents: 1 2effort 0.99Our capture
Inkling15 Jul 2026 · Thinking Machines74.1%thinkingmachines/Inkling model cardChecked A person checked this number against its source on 29 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.effort 0.99; another report, not chartedOur capture
GPT-5.523 Apr 2026 · OpenAI75.3%Introducing GPT-5.5Corroborated The same number is printed in 1 other lab document: 1Scale AI results after its April 2026 updateOur capture
Ripple-65B Preview4 Nov 2026 · Ricursive Intelligence71.67%ricursiveAI/Ripple-65B-Preview model cardChecked A person checked this number against its source on 7 Nov 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.OpenCode harness; maximum effortOur capture
GPT-5.45 Mar 2026 · OpenAI67.2%Introducing GPT-5.4Corroborated The same number is printed in 1 other lab document: 1xhigh reasoning effortOur capture
GPT-5.45 Mar 2026 · OpenAI70.6%Introducing GPT-5.5Checked A person checked this number against its source on 29 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.Scale AI results after its April 2026 update; reported later; another report, not chartedOur capture
Claude Opus 4.65 Feb 2026 · Anthropic62.7%Claude Opus 4.6Corroborated The same number is printed in 1 other lab document: 1high effort
Claude Opus 4.65 Feb 2026 · Anthropic59.5%Claude Opus 4.6Corroborated The same number is printed in 3 other lab documents: 1 2 3max effort; another report, not chartedSource image
Claude Opus 4.524 Nov 2025 · Anthropic62.3%Introducing Claude Opus 4.5Corroborated The same number is printed in 5 other lab documents. The first 3 are linked: 1 2 3no extended thinking; 200K contextSource image
Claude Sonnet 4.617 Feb 2026 · Anthropic61.3%Introducing Sonnet 4.6Corroborated The same number is printed in 1 other lab document: 1max effortSource image
GPT-5.211 Dec 2025 · OpenAI60.6%Introducing GPT-5.2Corroborated The same number is printed in 1 other lab document: 1xhigh reasoning effortOur capture
GPT-5.4 mini17 Mar 2026 · OpenAI57.7%Introducing GPT-5.4 mini and nanoChecked A person checked this number against its source on 29 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.xhigh reasoning effortOur capture
GPT-5.4 nano17 Mar 2026 · OpenAI56.1%Introducing GPT-5.4 mini and nanoChecked A person checked this number against its source on 29 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.xhigh reasoning effortOur capture
GPT-5 mini7 Aug 2025 · OpenAI47.6%Introducing GPT-5.4 mini and nanoChecked A person checked this number against its source on 7 Nov 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.high reasoning effort, the highest available; reported laterOur capture
GPT-5.112 Nov 2025 · OpenAI44.5%Introducing GPT-5.2Checked A person checked this number against its source on 29 Oct 2026. This confirms the number matches what the lab published. It doesn't confirm that the score was reproduced.high reasoning effort; reported laterOur capture

All benchmarks