Claude Haiku 4.5vsGrok 2.5
Claude Haiku 4.5 | Grok 2.5 | |
|---|---|---|
| API pricingUSD per 1M tokens · lower wins | ||
Input priceWhat you pay for everything you send the model — your question, plus any documents or earlier conversation you include with it. | $1.00 | — |
Output priceWhat you pay for the text the model writes back. It is normally the dearer half: producing an answer costs more than reading one. | $5.00 | — |
Cached input priceA reduced rate for text you send over and over. If every request starts with the same instructions or the same document, the provider keeps a copy ready and charges less to read it again. | $0.10 | — |
Cheapest inputLowest input rate across third-party providers, excluding the lab itself. The cheapest endpoint may run a quantised build or a shorter context — see "Available from" on the model page. | $1.00Amazon Bedrock | — |
Cheapest outputLowest output rate across third-party providers, excluding the lab itself. May come from a different provider than the cheapest input. | $5.00Amazon Bedrock | — |
| BenchmarksPublished by one model only | ||
BullshitBench v2Nonsense detection — Given a confidently-worded but nonsensical prompt, does the AI spot that it makes no sense and push back — instead of playing along and inventing an answer? The score is how often it clearly called out the nonsense. Higher is better. | 77% | — |
ProgramBenchProgram reconstruction — The AI receives a working program and its documentation, then builds a replacement from scratch without the original source code, internet access or decompilation. The score is the percentage of 200 programs that pass every behavioral test. We record each model's best published mini-SWE-agent result, including higher reasoning efforts where available. Partial test-pass rates and almost-solved programs do not count toward this score. Equal scores share a rank here; the official board also uses partial progress to break ties. Higher is better. | 0% | — |
SWE-Bench VerifiedCoding — Real coding tasks pulled from open-source projects — the AI has to find and fix actual bugs. A human-checked version of the original SWE-Bench. Higher is better. | 73.3% | — |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 73% | — |
| Overview | ||
| Company | Anthropic | SpaceXAI |
| Release date | Oct 15 2025 | Aug 23 2025 |
| Access | Proprietary | Proprietary |
Other comparisons
Claude Haiku 4.5vsGPT-6 AstraGrok 2.5vsGPT-6 AstraClaude Haiku 4.5vsGemini 3.8 FlashGrok 2.5vsGemini 3.8 FlashClaude Haiku 4.5vsMuse Spark 1.3Grok 2.5vsMuse Spark 1.3Claude Haiku 4.5vsDeepSeek-V4.1-FlashGrok 2.5vsDeepSeek-V4.1-FlashClaude Haiku 4.5vsMistral Medium 3.5Grok 2.5vsMistral Medium 3.5Claude Haiku 4.5vsKimi K3Grok 2.5vsKimi K3
Frequently asked questions
Claude Haiku 4.5 and Grok 2.5 don't publish scores on any of the same benchmarks, so there's no direct head-to-head comparison. Only Claude Haiku 4.5 has a verified first-party API price: $1.00 per million input tokens and $5.00 per million output tokens. No pay-as-you-go API rate is tracked for Grok 2.5. Grok 2.5 shipped 53 days before Claude Haiku 4.5, so benchmark comparisons should account for the intervening progress.
Published specifications for these two models are limited — see each model page for the latest details.
Direct benchmark comparisons are unavailable — Claude Haiku 4.5 and Grok 2.5 don't publish scores on any of the same benchmarks.