Recent studyBlind TTS Elo is live. Compare two anonymous voice samples, vote after listening, and help separate real preference signal from noise.Vote in the study ->
Codesota · Models · Step-3.5-Flash7 results · 7 benchmarks
Model card

Step-3.5-Flash.

unknown
§ 02 · Benchmarks

Every benchmark Step-3.5-Flash has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01AIME 2025Reasoning · Mathematical Reasoningaccuracy97.3%#2/22source ↗
02Tau2-BenchAgentic AI · Tool Useaccuracy88.2%#2/11source ↗
03SWE-Bench VerifiedComputer Code · Code Generationaccuracy74.4%#10/22source ↗
04BrowseCompNatural Language Processing · Question Answeringaccuracy51.6%#12/16source ↗
05HLEReasoning · Multi-step Reasoningaccuracy23.1%#18/36source ↗
06GPQA DiamondReasoning · Multi-step Reasoningaccuracy83.5%#27/74source ↗
07MMLU-ProReasoning · Commonsense Reasoningaccuracy84.4%#31/73source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Strengths by area

Where Step-3.5-Flash actually performs.

Agentic AI
1
benchmark
avg rank #2.0
Computer Code
1
benchmark
avg rank #10.0
Natural Language Processing
1
benchmark
avg rank #12.0
Reasoning
4
benchmarks
avg rank #19.5
§ 04 · Papers

1 paper with results for Step-3.5-Flash.

  1. 2026-02-11· 7 results

    Step 3.5 Flash: Open Frontier-Level Intelligence with 11B Active Parameters

§ 06 · Sources & freshness

Where these numbers come from.

pwc-dump
7
results
0 of 7 rows marked verified.