Terminal-Bench Math Modeling
Agents use a terminal to solve 18 mathematical modeling problems and submit numerical answers. Feasibility checks and numerical scoring compare answers with reference solutions.
The current board uses Multi-O against 99 human papers. The paper and repository also describe different one-reference scoring rules. Their results are not interchangeable.
- Made by
- TB-MathModel authors
- Released
- November 2026
- Kind
- Scientific code
- Level
- Competition
- Scored
- Checked automatically
- Answers
- Agent task
- Tasks
- All public
- Measures
- Capability
- Subject
- Mathematics
- Language
- English
- Interaction
- Agentic
- Environment
- Terminal
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
- GPT-5.6 Sol #14 on the capability index
- Last model added
- September 2026
- Scores refreshed
- November 20th, 2026
In the authors’ words
Mathematical modeling turns an open-ended real-world problem into assumptions, a model, a solution, and a defended recommendation, and its answers matter only if they are both close to a credible reference and physically feasible; existing benchmarks either score solver fragments or grade reports with an LLM judge, leaving those numbers unchecked. We introduce TB-MathModel (the Terminal-style Mathematical Modeling Bench), the first benchmark to make complete contest-style modeling problems agentic, verifiable, and terminal at once, over 18 real CUMCM and MCM problems (2023–2025). Each task ships the verbatim statement, official attachments, and figures in a containerized terminal; the agent returns one JSON artifact, scored on the numeric core of the final question against a reproduction of the Outstanding-award paper. Two design choices distinguish it: direction-aware, one-sided scoring, which never penalises beating the anchor, and two-layer feasibility verification—a programmatic gate over six decidable invariant forms plus 624 Lean 4 obligations—with a hard gate that zeroes infeasible answers. Each task is scored against a panel of human outstanding papers rather than one, so a model must be close to many accepted answers; across seven multimodal entries (three seeds each) the strongest reach a multi-reference O-Eval of 64.58%, while the $2.54 open model holds 59.81% at 1/100 the cost. TB-MathModel turns the numeric core of modeling into a reproducible, machine-checkable signal for measuring long-horizon modeling competence and for collecting agent trajectories. The source code is available at https://github.com/wxj630/terminal-bench-math-modeling.
GPT-5.5 leads with 64.3 among models released by September 2026 GPT-5.5 leads with 64.3 among models released by September 2026
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. 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 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- Claude Opus 5.5
- and 15 more
- 1 GPT-5.5 Multimodal / xhigh 64.3
- 2 Kimi K3 Text-only 64.2
- 3
Tencent Hy4 preview Text-only 63.9
- 4
Qwen 3.8 27B Text-only / thinking 62.2
- 5
DeepSeek V4.1 Flash Multimodal 62.2
- 6 GPT-5.6 Sol Text-only / high 60.6
- 7 v4 flash baseline Text-only 60.2
- 8 GLM-5.3 Text-only 59.3
- 9 v4 pro Text-only 58.4
- 10 GLM-5.3-Flash (ox-alpha) Text-only 55.1
- 11
Gemini 3.7 Flash Text-only / high 54.9
- 12 Qwen3.8 Flash (Bailian) Text-only 51.7
Source: Official Terminal-Bench Math Modeling leaderboard
- 1 GPT-5.5 Multimodal / xhigh 64.3
- 2 Kimi K3 Text-only 64.2
- 3
Tencent Hy4 preview Text-only 63.9
- 4
Qwen 3.8 27B Text-only / thinking 62.2
- 5
DeepSeek V4.1 Flash Multimodal 62.2
- 6 GPT-5.6 Sol Text-only / high 60.6
- 7 GLM-5.3 Text-only 59.3
- 8
Gemini 3.7 Flash Text-only / high 54.9
Source: Official Terminal-Bench Math Modeling leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.5 | 64.3 | Multimodal / xhigh, codex | Apr 2026 | Official leaderboard |
| 2 | Kimi K3 | 64.2 | Text-only | Jul 2026 | Official leaderboard |
| 3 | | 63.9 | Text-only | Aug 2026 | Official leaderboard |
| 4 | | 62.2 | Text-only / thinking | Aug 2026 | Official leaderboard |
| 5 | | 62.2 | Multimodal, opencode | Sep 2026 | Official leaderboard |
| 6 | GPT-5.6 Sol | 60.6 | Text-only / high | Jul 2026 | Official leaderboard |
| 7 | v4 flash baseline | 60.2 | Text-only | Official leaderboard | |
| 8 | GLM-5.3 | 59.3 | Text-only | Aug 2026 | Official leaderboard |
| 9 | v4 pro | 58.4 | Text-only | Official leaderboard | |
| 10 | GLM-5.3-Flash (ox-alpha) | 55.1 | Text-only | Official leaderboard | |
| 11 | | 54.9 | Text-only / high | Aug 2026 | Official leaderboard |
| 12 | Qwen3.8 Flash (Bailian) | 51.7 | Text-only | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-5.5 | 64.3 | Multimodal / xhigh, codex | Apr 2026 | Official leaderboard |
| 2 | Kimi K3 | 64.2 | Text-only | Jul 2026 | Official leaderboard |
| 3 | | 63.9 | Text-only | Aug 2026 | Official leaderboard |
| 4 | | 62.2 | Text-only / thinking | Aug 2026 | Official leaderboard |
| 5 | | 62.2 | Multimodal, opencode | Sep 2026 | Official leaderboard |
| 6 | GPT-5.6 Sol | 60.6 | Text-only / high | Jul 2026 | Official leaderboard |
| 7 | GLM-5.3 | 59.3 | Text-only | Aug 2026 | Official leaderboard |
| 8 | | 54.9 | Text-only / high | Aug 2026 | Official leaderboard |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official Terminal-Bench Math Modeling leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official Terminal-Bench Math Modeling 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
- 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
All 9 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (12 entries).
- Pass Models scored 8 models we track, plus 4 other entries on its leaderboard.
- Pass Newest score The newest is from September 2026, for DeepSeek V4.1 Flash.
- Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.6 Sol (#14).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on wxj630.github.io: 10 rows for models we track and 4 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.