SI Index Data as of Oct 9, 2026

Qwen3 235B-A22B Instruct 2507: Price, Context, Benchmarks, and Release Details

Alibaba / Qwen provisional listing open weights · first seen Oct 8, 2026 · score computed Oct 9, 2026, 01:09 UTC
42.4
SI Score (method si-v3-retained-evidence-2)
48% confidence 48 percent, Low confidence — 2 of 7 expected sources in
Coverage 38% of expected source weight · 100% confidence at 80% coverage

Pillar breakdown

Coding (weight 40 percent) —
Math (weight 15 percent) —
Preference (weight 15 percent) 75.0
Reasoning (weight 30 percent) 7.5

Pillar weights: reasoning 30% · math 15% · coding 40% · preference 15%. Benchmark results use fixed 0–100 scales before averaging; incomplete evidence is shrunk toward 50.

Facts

Input price / 1M
$0.23Alibaba Model Studio pricingOfficial Alibaba Model Studio Global USD on-demand API; standard tier; Non-Thinking mode only. Output uses non-thinking rate when both modes are available; thinking-only products use their thinking rate. Cache, Batch, free quotas and regional rates excluded Retrieved Oct 9, 2026 · factual citation
Open source ↗
Output price / 1M
$0.92Alibaba Model Studio pricingOfficial Alibaba Model Studio Global USD on-demand API; standard tier; Non-Thinking mode only. Output uses non-thinking rate when both modes are available; thinking-only products use their thinking rate. Cache, Batch, free quotas and regional rates excluded Retrieved Oct 9, 2026 · factual citation
Open source ↗
Context window
262Kmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
Max output
16.4Kmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
Released
Jul 21, 2025models.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
Open weights
Yesmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
License
Apache 2.0models.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗
Input modalities
textmodels.devPublished source fact Retrieved Oct 9, 2026 · MIT
Open source ↗

Benchmark results

5 results
Benchmark Raw result Normalized (0–100) Pillar
arc agi v1 public eval 17%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] qwen3-235b-a22b-instruct-2507Published Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 		17.0 reasoning
arc agi v1 semi private 11%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] qwen3-235b-a22b-instruct-2507Published Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 		11.0 reasoning
arc agi v2 public eval 0.8%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] qwen3-235b-a22b-instruct-2507Published Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 		0.8 reasoning
arc agi v2 semi private 1.3%ARC PrizeARC Prize steward-published result; exact edition/subset; publication time is HTTP Last-Modified of the aggregate export, not the evaluation date. [variant] qwen3-235b-a22b-instruct-2507Published Oct 6, 2026 Retrieved Oct 9, 2026 · factual citation
Open source ↗ 		1.3 reasoning
lmarena text 1419.3 eloLMArena / ArenaPublished source fact [variant] text / overallPublished Oct 2, 2026 Retrieved Oct 9, 2026 · CC-BY-4.0
Open source ↗ 		75.0 preference

Normalization uses fixed absolute 0–100 scales for each unit, independently of other models that have a result on each benchmark. Coverage and evidence breadth still affect the composite; compare the evaluation conditions before reading a small score gap as decisive. “lab-reported” marks the provider's own published figure.

Sources: in and pending

in Reported (2)

  • ARC Prize · arrived Oct 8, 2026
  • LMArena / Arena · arrived Oct 8, 2026

pending Awaiting (5)

  • Epoch AI Benchmarking · carries 25% of expected weight
  • Humanity’s Last Exam · carries 13% of expected weight
  • Official model cards via models.dev · carries 4% of expected weight
  • LiveBench · carries 13% of expected weight
  • Terminal-Bench · carries 6% of expected weight

The confidence % rises as pending sources publish. Some sources never cover some models — that is why 100% confidence arrives at 80% of expected weight, not at full coverage.