Model comparison
Claude Haiku 4.5 vs GPT-5.2 Codex
GPT-5.2 Codex is the stronger model overall, scoring 42.2 to 39.3 on the Noometry Index. Claude Haiku 4.5 costs 2.4× less per token, which makes it the better buy when GPT-5.2 Codex's lead doesn't matter for your workload.
Last verified . 5 shared benchmarks.
Summary
- They share 5 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 0 categories and GPT-5.2 Codex in 2 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where GPT-5.2 Codex leads 40.7 to 33.5.
- The biggest single-benchmark swing is Terminal-Bench: 35.5% for Claude Haiku 4.5 and 66.5% for GPT-5.2 Codex.
- Claude Haiku 4.5 is cheaper at $1 / $5 per million input/output tokens, against $1.75 / $14 for GPT-5.2 Codex.
- GPT-5.2 Codex accepts more context: 400K tokens versus 200K.
Side by side
| Claude Haiku 4.5 | GPT-5.2 Codex | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 39.3 | 42.2 |
| Released | 2025-10-15 | 2025-12-18 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 400K |
| Max output | 64K | 128K |
| Input $ / M tokens | $1 | $1.75 |
| Output $ / M tokens | $5 | $14 |
| Results tracked | 53 | 5 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.2 Codex leads
Claude Haiku 4.5: 43.9 (#71), GPT-5.2 Codex: 45.4 (#66)
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| SWE-bench Verified (bash only) | 66.6% | 72.8% |
| LMArena WebDev | 1330 | 1339 |
| SWE-bench Multilingual | 64.7% | 66.3% |
| ALE-Bench | 653.48 | 1,300 |
| SciCode | 43.3% | — |
| WeirdML | 45.4% | — |
| LMArena Coding | 1454 | — |
Agentic & Tool Use GPT-5.2 Codex leads
Claude Haiku 4.5: 33.5 (#50), GPT-5.2 Codex: 40.7 (#20)
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| Terminal-Bench | 35.5% | 66.5% |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| Vending-Bench 2 | 458.89 | — |
Reasoning Not comparable
Claude Haiku 4.5: 15.1 (#310), GPT-5.2 Codex: —
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| ARC-AGI-2 | 4% | — |
| NYT Connections (extended) | 14.3% | — |
| ARC-AGI-1 | 47.7% | — |
| CritPt | 0% | — |
| Chess Puzzles | 8% | — |
| LMArena Hard Prompts | 1420 | — |
| DTBench | 73.6% | — |
| LMCA | 30.9% | — |
| Epoch Capabilities Index | 142.42 | — |
| ForecastBench | 61.4 | — |
Math Not comparable
Claude Haiku 4.5: 43.0 (#69), GPT-5.2 Codex: —
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | — |
| Omni-MATH | 56.1% | — |
| LMArena Math | 1395 | — |
| MATH Level 5 | 96.4% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge Not comparable
Claude Haiku 4.5: 37.6 (#146), GPT-5.2 Codex: —
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| GPQA Diamond | 71.2% | — |
| SimpleQA Verified | 13.2% | — |
| MMLU-Pro | 77.7% | — |
| Vectara Hallucination Rate | 9.8% | — |
| GPQA (HELM) | 60.5% | — |
| LMArena Expert | 1443 | — |
Multimodal Not comparable
Claude Haiku 4.5: 27.0 (#111), GPT-5.2 Codex: —
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| Blueprint-Bench 2 | 0% | — |
| LMArena Document | 1420 | — |
Multilingual Not comparable
Claude Haiku 4.5: 50.1 (#124), GPT-5.2 Codex: —
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| LMArena Non-English | 1376 | — |
| LMArena Chinese | 1415 | — |
| LMArena French | 1408 | — |
| LMArena German | 1377 | — |
| LMArena Japanese | 1339 | — |
| LMArena Korean | 1346 | — |
| LMArena Russian | 1380 | — |
| LMArena Spanish | 1420 | — |
Instruction Following Not comparable
Claude Haiku 4.5: 71.6 (#144), GPT-5.2 Codex: —
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| IFEval | 80.1% | — |
| LMArena Instruction Following | 1414 | — |
Long Context Not comparable
Claude Haiku 4.5: 43.6 (#86), GPT-5.2 Codex: —
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| LMArena Longer Query | 1427 | — |
Writing & Preference Not comparable
Claude Haiku 4.5: 58.1 (#119), GPT-5.2 Codex: —
| Benchmark | Claude Haiku 4.5 | GPT-5.2 Codex |
|---|---|---|
| LMArena Text | 1396 | — |
| LMArena Creative Writing | 1371 | — |
| WildBench | 83.9% | — |
| EQ-Bench 4 | 1064 | — |
| LMArena Multi-Turn | 1409 | — |
Frequently asked questions
Is Claude Haiku 4.5 better than GPT-5.2 Codex?
GPT-5.2 Codex is the stronger model overall, scoring 42.2 to 39.3 on the Noometry Index. Claude Haiku 4.5 costs 2.4× less per token, which makes it the better buy when GPT-5.2 Codex's lead doesn't matter for your workload.
Which is cheaper, Claude Haiku 4.5 or GPT-5.2 Codex?
Claude Haiku 4.5 is cheaper. It lists at $1 per million input tokens and $5 per million output tokens; GPT-5.2 Codex lists at $1.75 and $14.
Is Claude Haiku 4.5 or GPT-5.2 Codex better for coding?
GPT-5.2 Codex scores higher on coding benchmarks: 45.4 versus 43.9 in the Noometry coding category.
Which has the bigger context window?
GPT-5.2 Codex does, with 400K tokens against 200K.
How many benchmarks do Claude Haiku 4.5 and GPT-5.2 Codex share?
5 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and GPT-5.2 Codex has 5.