AIEvals

Benchmarks

GraphWalks (256k to 1M)

Breadth-first graph traversal over very long inputs, scored with F1.

Long contextReported by Google DeepMindVendor-reported onlyReported by Google DeepMind

How it is run

Google self-computes all model results on 200 GraphWalks problems with context lengths between 256k and 1M tokens.

What it cannot tell you

BFS F1 is not a binary task pass rate. The up-to-128k subset has different items and its own page; these scores do not establish a public API context limit.

Why there is no independent table

No independent suite in this dataset runs GraphWalks (256k to 1M). The numbers below are the ones model publishers report about their own models, which is a different kind of evidence: each publisher chooses its own harness and effort level, so a row is readable on its own and not against the other rows.

Read this with

Other versions of the same evaluation. Versions are not comparable: the task set changes between them.

As reported by each publisher

Kept separate per publisher, because the settings are not shared.

Google DeepMind

Gemini 4 Argon: model evaluation · October 2026

RankModelScore
1Gemini 4 Argon84.2%
2GPT-6 Astra71.8%
4Claude Fable 5.165.0%
3Claude Opus 5.566.8%

Ranks compare reported scores in the same published row; ties share a rank.

GraphWalks (256k to 1M, BFS F1). 200 problems with input contexts between 256k and 1M tokens; BFS F1.

Page 5 of Google DeepMind’s model evaluation, read 1 October 2026. Gemini runs use the API at the highest thinking setting, generally pass@1 without majority voting or parallel test-time compute; smaller benchmarks average multiple trials. Other models use their highest available published reasoning results. Google quotes Vals AI for knowledge-work and Vibe Code results, Zapier’s private-set AutomationBench leaderboard, Proximal for FrontierSWE v2, Surge AI for RiemannBench and no-tools Chartography, and the CWE-bench v1 leaderboard. Argon’s DeepSWE uses mini-swe; Terminal-Bench 4.0 is self-computed for Argon and quoted from the author board for other models. PostTrainBench v1.1 uses OpenCode with one H100 and a 10-hour budget for all models. Terminal-Bench Science uses a 6x verifier timeout for Argon. LABBench 2 uses terminal, bioinformatics tools, Python, R and internet access for all models. GraphWalks uses BFS F1 on two separate subsets: 650 items up to 128k tokens and 200 items from 256k to 1M. Agent’s Last Exam uses ALE-Claw, a 5-hour window and binary pass rate, with safety filters enabled for Argon. OSWorld 2.0 uses offline partial reward, the best of three single-attempt runs for Argon, safety filters, 1080p, 500 steps, Gemini CUA with parallel batch tools and compaction, screenshot-only observation and the 08.08 patch; Claude’s combined online/offline results are omitted. LVBench runs without tools, using 1 FPS for Gemini, 800 frames for Astra, 300 for Fable and 600 for Opus. Missing cells remain blank. These are the report’s published figures, preserved separately from this site’s independent source snapshots.

storage.googleapis.com/deepmind-media/gemini/gemini_4_argon_model_evaluation.pdf

Sources

Every page used for this benchmark, read on 17 October 2026.