Benchmark Almanac

BIRD-Python

Qwen3-Max 73.2% leads since January 2026
Not in model cards

Models write pandas code to answer questions about tables. A language model compares the executed output with a reference result.

Optum AI tested nine models on the original questions alongside two rewrites. Its analysis excluded 18 of the 1,534 source tasks.

Released
January 2026
Kind
Data science and SQL
Level
University
Scored
Graded by a model
Answers
Code
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Tables → Code
Interaction
Single turn
Environment
Code sandbox
Items from
Another benchmark
Refresh
One fixed set
Run by
Anyone
Based on
Bird-SQL
Strongest model tested
Qwen3-Max #111 on the capability index
Last model added
September 2025
Scores refreshed
November 20th, 2026
14
Entries in its paper, 8 models we track

Qwen3-Max leads with 73.2% among models released by September 2025 Qwen3-Max leads with 73.2% among models released by September 2025

Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in September 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Qwen3-Max Logical clarifications 73.2%
  2. 2 Qwen3-32B Logical clarifications 72.5%
  3. 3 Qwen3-7B Logical clarifications 71.2%
  4. 4 DeepSeek-R1 (snapshot unspecified) Standard 62.5%
  5. 5 Qwen3-Coder-Plus Standard 61.6%
  6. 6 Qwen-Max (snapshot unspecified) Standard 60.8%
  7. 7 Qwen2.5-Coder-32B Standard 56.6%
  8. 8 Qwen2.5-Coder-14B Standard 54.3%
  9. 9 Qwen3-14B Standard 48%
  10. 10 Qwen3-Coder-30B Standard 45.8%
  11. 11 Qwen2.5-Coder (7B) Standard 36%
  12. 12 TableGPT2-7B Standard 19.8%
  13. 13 AutoCoder-7B Standard 5.45%
  14. 14 StarCoder 2 7B Standard 1.12%

Source: The BIRD-Python paper

#EntryScoreSettingDateSource
1 Qwen3-Max 73.2% Logical clarifications Sep 2025 Official leaderboard
2 Qwen3-32B 72.5% Logical clarifications Apr 2025 Official leaderboard
3 Qwen3-7B 71.2% Logical clarifications Official leaderboard
4 DeepSeek-R1 (snapshot unspecified) 62.5% Standard Official leaderboard
5 Qwen3-Coder-Plus 61.6% Standard Sep 2025 Official leaderboard
6 Qwen-Max (snapshot unspecified) 60.8% Standard Official leaderboard
7 Qwen2.5-Coder-32B 56.6% Standard Sep 2024 Official leaderboard
8 Qwen2.5-Coder-14B 54.3% Standard Sep 2024 Official leaderboard
9 Qwen3-14B 48% Standard Apr 2025 Official leaderboard
10 Qwen3-Coder-30B 45.8% Standard Official leaderboard
11 Qwen2.5-Coder (7B) 36% Standard Sep 2024 Official leaderboard
12 TableGPT2-7B 19.8% Standard Official leaderboard
13 AutoCoder-7B 5.45% Standard Official leaderboard
14 StarCoder 2 7B 1.12% Standard Feb 2024 Official leaderboard
#ModelScoreSettingDateSource
1 Qwen3-Max 73.2% Logical clarifications Sep 2025 Official leaderboard
2 Qwen3-32B 72.5% Logical clarifications Apr 2025 Official leaderboard
3 Qwen3-Coder-Plus 61.6% Standard Sep 2025 Official leaderboard
4 Qwen2.5-Coder-32B 56.6% Standard Sep 2024 Official leaderboard
5 Qwen2.5-Coder-14B 54.3% Standard Sep 2024 Official leaderboard
6 Qwen3-14B 48% Standard Apr 2025 Official leaderboard
7 Qwen2.5-Coder (7B) 36% Standard Sep 2024 Official leaderboard
8 StarCoder 2 7B 1.12% Standard Feb 2024 Official leaderboard

Rankings by part

We read 3 parts from the benchmark’s paper: Simple, Moderate and Challenging. Each ranks models on one part of the test. Qwen3-Max leads 2 of them, and Qwen3-32B leads the other one. Arrows show how far a model moved from its overall rank among the models we track.

Simple

8 models we track have a score here.

  1. 1 Qwen3-32B ↑1 78.2%
  2. 2 Qwen3-Max ↓1 77.8%
  3. 3 Qwen3-Coder-Plus 66%
  4. 4 Qwen2.5-Coder-32B 62.2%
  5. 5 Qwen2.5-Coder-14B 60.8%
  6. 6 Qwen3-14B 54.7%
  7. 7 Qwen2.5-Coder (7B) 41.9%
  8. 8 StarCoder 2 7B 1.95%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Qwen
Benchmark paper Not tested since Jan 2026
Qwen3-Max 73.2%

Source: The BIRD-Python paper

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2024 2025 Near the ceiling (85%) Qwen3-Max Qwen3-32B Qwen2.5-Coder-32B StarCoder 2 7B 0% 20% 40% 60% 80% 100% 2024 2025 Near the ceiling (85%) Qwen3-Max Qwen2.5-Coder-32B

Source: Official BIRD-Python leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Coding

All 523 in Coding
  • SWE-bench Verified In use 500 human-validated GitHub issues from 12 Python repos. Agent writes a patch that must pass hidden tests. Claude Opus 4.7 83.5%
  • LiveCodeBench v6 In use Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. DeepSeek-V4-Pro 93.5%
  • SWE-bench Multilingual In use 300 GitHub-issue tasks across nine programming languages in SWE-bench format. Patch graded by repository tests. Claude Fable 5.2 96.6%
  • DeepSWE In use 113 original long-horizon engineering tasks across 91 repos and five languages, written from scratch. Hand-written verifiers accept alternative solutions. GPT-6 Astra 74.1%
  • SciCode In use Scientist-curated research coding problems split into subproblems across natural sciences. Code checked against test cases. Seed2.1 Pro 59.8%
  • Codeforces rating In use Model solves recent Codeforces contest problems. An Elo-style rating is computed from simulated contest performance. DeepSeek V4.1 Flash 3,471

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 14 months ago, in September 2025, for Qwen3-Max. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the benchmark’s paper (14 entries).
  • Pass Models scored 8 models we track, plus 6 other entries on its leaderboard.
  • Warning Newest score The newest came 14 months ago, in September 2025, for Qwen3-Max.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Qwen3-Max (#111).
  • Pass Headroom The best is 73.2%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on arxiv.org: 10 rows for models we track and 6 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check