Claude Sonnet 4.5
Anthropic · released Sep 29, 2025
- Type
- Proprietary
- Input
- $3.00
- Output
- $15.00
- Cached
- $0.30
List price · per 1M tokens · Anthropic at 1M context · source ↗
Our take
Written Aug 2, 2026Claude Sonnet 4.5 is Anthropic's proprietary text-and-image model that can handle up to one million tokens in a single request. It has the widest benchmark coverage we track on the Arena leaderboard, with coding as its standout measured skill.
Pick this for long-document analysis at one-million-token scale, coding tasks where its Arena Coding score is strongest, or hard-prompt and instruction-following workloads. Use it when you want provider flexibility with five options available. Skip it if you need predictable throughput on every endpoint, if maths consistency matters more than coding strength, or if you want to self-host rather than use an API.
The case for it
- Widest benchmark coverage in the Arena leaderboard set we track, with ten distinct categories measured across six skill areas.
- Strongest measured skill is coding, at 64 points above its own overall text score.
- Exceptional context length for long-document work, with a one-million-token request limit.
- Consistent base-tier pricing across most offers and four providers.
The case against it
- Throughput is unpredictable and can collapse on some endpoints, with a more than twelvefold spread between the slowest and fastest measured provider.
- Maths performance shows significant run-to-run variance, the widest of any duplicated benchmark we hold.
- Web development coding lags general coding by over 120 points on the same evaluation framework.
How good is it?
IntelligencePuzzles, maths, exam questions
Arena Text (overall)32nd of 143 · 1455
Also on this board: 1455.6 via Thinking 32k (Aug 2, 2026). Read the pair, not the higher one.
CodingWriting and fixing code on its own
Arena Coding22nd of 143 · 1513.1
Also on this board: 1519.6 via Thinking 32k (Aug 2, 2026). Read the pair, not the higher one.
AgenticPlanning, calling tools, staying on task
Claude Sonnet 4.5 is not on Arena Agent (IPS), which is where the rating would come from, so there is no rating here. It is on SWE-bench Verified, in 4th of 39 with 74.8.
WritingWe do not rate this
Two boards come close and neither tests writing: Arena Creative Writing asks people which of two replies they prefer, and LiveBench Language tests whether a model understood a passage. So we show where Claude Sonnet 4.5 placed and give it no mark out of five.
These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done, which is why they get no rating.
Every published score for this model8 scoresEvery figure we hold, from 8 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Or rent it from someone else
The lab is also the cheapest we hold. The strip above and this offer are the same one, so nothing on this page undercuts Anthropic on 1M of context.
- per 1M tokens
- $3.00 in / $15.00 out
- Context served
- 1M
- Throughput
- ~26 tok/s
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| Amazon Bedrock | $3.00 / $15.00 | 1M | 36 tok/s | No | No | Confirmed |
| Microsoft Azure AIglobal | $3.00 / $15.00 | 200K | 38 tok/s | No | No | Unknown |
| Google Vertex AIglobal | $3.00 / $15.00 | 1M | 36 tok/s | No | No | Confirmed |
| Anthropicclaude-on-aws | $3.00 / $15.00 | 1M | 26 tok/s | No | Yes30 days | Unknown |
| OpenRouter | $3.00 / $15.00 | 1M | not measured | Unknown | Unknown | Unknown |
| Anthropic | $3.00 / $15.00 | 200K64K out | 37 tok/s | No | Yes30 days | Unknown |
| Google Vertex AIus-east5 | $3.30 / $16.50 | 1M | 27 tok/s | No | No | Confirmed |
| Amazon Bedrockeu-west-1 | $3.30 / $16.50 | 1M | 3 tok/s | No | No | Confirmed |
Amazon Bedrock
Microsoft Azure AI
Google Vertex AI
OpenRouterAcross the 8 listings we hold: 7 say they do not train on prompts, 0 say they do and 1 do not say. 4 appear in the zero-retention registry we check; the rest are unknown to us rather than confirmed either way.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
- ✓
- Supported
- ✗
- Not supported
- Not published
- host gave no parameter list
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| Amazon Bedrock | ✓ | ✓ | ✓ |
| Microsoft Azure AIglobal | ✓ | ✓ | ✓ |
| Google Vertex AIglobal | ✓ | ✗ | ✗ |
| Anthropicclaude-on-aws | ✓ | ✓ | ✓ |
| OpenRouter | ✓ | ✓ | ✓ |
| Anthropic | ✓ | ✓ | ✓ |
| Google Vertex AIus-east5 | ✓ | ✗ | ✗ |
| Amazon Bedrockeu-west-1 | ✓ | ✓ | ✓ |
Tool calling: 8 of 8 listings say yes. JSON output: 6 of 8 listings say yes, 2 say no. Strict schema: 6 of 8 listings say yes, 2 say no.
Models people weigh against Claude Sonnet 4.5
When we formed this view
Dates behind this page
Prices last checked 2d ago
What we do not know about this model yet
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 1 of 8 listings do not say whether they train on prompts.
Licence and identifiers
What the licence allowsWe hold no licence record for this model. Inside are the identifiers you need to pull it — its Hugging Face repo where we have one, our slug and a machine-readable card.
Licence
Commercial API terms. We hold no licence record for this model, so there is nothing to summarise here.
Identifiers
- Modality record
- text+image+file->text
- Catalogue slug
- anthropic-claude-sonnet-4-5