Leaderboard · Compare · GPT-5 mini vs Claude 3.5 Haiku · Updated
GPT-5 mini vs Claude 3.5 Haiku
GPT-5 mini edges out Claude 3.5 Haiku on the composite (77.0 vs 56.2). The gap is meaningful but not decisive — see the per-benchmark breakdown below.
At a glance
| Spec | GPT-5 mini | Claude 3.5 Haiku |
|---|---|---|
| Provider | OpenAI | Anthropic |
| Released | 2025-08 | 2024-11 |
| Tier | fast / cheap | fast / cheap |
| License | Closed | Closed |
| Context window | 400k | 200k |
| $ in / out (per 1M) | $0.25 / $2.00 | $0.80 / $4.00 |
Benchmark scoreboard
Higher is better on every benchmark. Δ shows GPT-5 mini − Claude 3.5 Haiku.
| Benchmark | GPT-5 mini | Claude 3.5 Haiku | Δ |
|---|---|---|---|
| Chatbot Arena Elo | 1370 | 1240 | +130 |
| MMLU-Pro | 80.1 | 65.0 | +15.1 |
| GPQA Diamond | 75.0 | 41.6 | +33.4 |
| MATH | 91.0 | 69.4 | +21.6 |
| HumanEval | 90.5 | 88.1 | +2.4 |
| SWE-Bench Verified | 60.5 | 40.6 | +19.9 |
Numbers compiled from provider technical reports and Chatbot Arena snapshots — see methodology.
OpenRouter routes GPT-5 mini, Claude 3.5 Haiku, and 100+ other LLMs behind a single API key — pay-as-you-go, no monthly minimum, fallback if a provider is down. Try OpenRouter → (affiliate · supports this site)
GPT-5 mini vs Claude 3.5 Haiku: where each one wins
GPT-5 mini is stronger on
- Arena
- MMLU-Pro
- GPQA
- MATH
- HumanEval
- SWE-Bench
Claude 3.5 Haiku is stronger on
No benchmarks where Claude 3.5 Haiku beats GPT-5 mini with comparable data.
Cost comparison
At 10M tokens/day (50/50 split), GPT-5 mini costs ~$11.25/day vs $24.00/day for Claude 3.5 Haiku — GPT-5 mini is the cheaper pick at this volume.
Verdict
GPT-5 mini edges out Claude 3.5 Haiku on the composite (77.0 vs 56.2). The gap is meaningful but not decisive — see the per-benchmark breakdown below.
If you can only pick one and your workload is unclear, route via OpenRouter and switch by request — same key, no lock-in.
Frequently asked questions
Which is better, GPT-5 mini or Claude 3.5 Haiku?
GPT-5 mini edges out Claude 3.5 Haiku on the composite (77.0 vs 56.2). The gap is meaningful but not decisive — see the per-benchmark breakdown below. GPT-5 mini wins on Arena, MMLU-Pro, GPQA, MATH, HumanEval, SWE-Bench; Claude 3.5 Haiku wins on no benchmarks.
What does GPT-5 mini cost compared to Claude 3.5 Haiku?
At 10M tokens/day (50/50 split), GPT-5 mini costs ~$11.25/day vs $24.00/day for Claude 3.5 Haiku — GPT-5 mini is the cheaper pick at this volume.
What is the context window of GPT-5 mini vs Claude 3.5 Haiku?
GPT-5 mini: 400k tokens. Claude 3.5 Haiku: 200k tokens. GPT-5 mini has the larger window — useful for long-document RAG and full-codebase prompting.
Is GPT-5 mini or Claude 3.5 Haiku open source?
GPT-5 mini: closed / proprietary. Claude 3.5 Haiku: closed / proprietary.
Can I try GPT-5 mini and Claude 3.5 Haiku on the same API key?
Yes — OpenRouter routes both models behind a single key, so you can A/B test GPT-5 mini against Claude 3.5 Haiku without juggling provider accounts.
Model deep-dives: GPT-5 mini · Claude 3.5 Haiku · Full leaderboard
Spotted out-of-date numbers? Open an issue — corrections usually ship within 24h.
Try GPT-5 mini and Claude 3.5 Haiku now
One API key, both models — switch between them per request and let real traffic pick the winner.