Claude Sonnet 5
Anthropic released 2026-06-30 proprietary
Muse Spark 1.2 is both better and cheaper.
It scores +1.5 higher and costs 50% less ($2.00/M against $4.00/M) on this workload — and it does everything this model does.
GLM 5.3 is cheaper still (46% less) but drops no image, file input.
The same model via batch is 50% cheaper ($2.00/M) — same weights, different latency.
Our take
editorial — not a measurementThe volume workhorse and probably the best-value proprietary model in this dataset. $2/$10 with a full 1M-token window and 128k max output undercuts Sonnet 4.6 ($3/$15) by a third while raising the ceiling. The pricing history matters for planning: $2/$10 shipped as introductory pricing through 2026-08-31 with a scheduled rise to $3/$15 on 2026-09-01, and Anthropic has now confirmed that increase will not happen — so this is the standard rate, not a countdown. Use it as the default for anything that does not specifically need Opus-class reasoning.
- Unverified in this take: Anthropic does publish a token-inflation figure, but it compares newer Claude models against EARLIER CLAUDE models — not against another vendor. The cross-vendor version of this claim is unsupported, so read any comparison to a non-Claude price with that in mind.
Strengths
- $2/$10 for a 1M context window and 128k output
- Introductory rate confirmed permanent — no September price rise
- Flat pricing across the full context window
- Adaptive thinking without the Opus price
Weaknesses
- Not competitive with Opus 5 on the hardest agentic tasks
- Uses the newer tokenizer, so ~30% token inflation vs Sonnet 4.6
- No independent LMArena or AA score captured for it in this pass
Reach for it when
- High-volume production traffic
- RAG and summarisation at scale
- Default choice when Opus is overkill
Avoid it if
- The workload is genuinely frontier-hard
- You need the absolute cheapest token price — open weights win there
Every price dimension
| Input | $2.00/M |
|---|---|
| Output | $10.00/M |
| Cached input | $0.200/M |
| Cache write | $2.50/M |
| Cache write 1h | $4.00/M |
| Web search | $0.01/call |
| Batch discount | 50% |
Independent scores
| Intelligence | 55.3 |
|---|---|
| Coding | 71.5 |
| Agentic | 49.7 |
| LMArena Elo | 1442.3high |
Sources are listed separately rather than averaged. Across the 62 models both have scored they correlate at r = 0.806 — close agreement overall, but four models rank very differently between them, and a blended score would hide exactly those.
Best rankings by task
- godotgamedev #4of 43 1268
- htmlslides #4of 34 1231
- agenticgamedev #5of 34 1236
- webapps #6of 56 1281
- fullstack #7of 59 1283
- python-pptxslides #7of 35 1249
- gamedev #9of 144 1329
- androidnative #9of 54 1249
Capability
| Context window | 1M |
|---|---|
| Max output | 128K |
| Input modes | text, image, file |
| Tool use | yes |
| Reasoning | optional |
| Open weights | no |
Provenance
| Price source | openrouter.ai |
|---|---|
| Fetched | 2026-08-24 |
| Quality data | verified |
| Cross-checked | vendor page |
Sonnet 5 is Anthropic's most capable Sonnet-class model, with frontier performance across coding, agents, and professional work. It supports adaptive thinking with selectable reasoning effort levels (low, medium, high, max,...