Provider Dossier · Frontier Lab
Anthropic
Long-context reasoning, strong safety posture and enterprise trust.
The Datum Index verdict
Benchmark of the field
Ranked No. 2 of 20 in our composite reputation index, and currently improving.
≈ 4.5 / 5 aggregate · 1,105 reviewer notes
Composite index · reviewed 10 July 2026
Findings
- Anthropic holds a Datum Index composite reputation of 90 / 100 (≈ 4.5 / 5), ranked No. 2 of 20.
- It is a frontier lab based in San Francisco, USA, founded 2021, offering proprietary model access.
- Its flagship offering is Claude family, through its own API.
- Indicative pricing is $3.00 / 1M tokens in / $15.00 / 1M tokens out (medium confidence).
- Reference throughput is ~55 tok/s with ~1300 ms time-to-first-token; stated uptime is 99.8%.
Anthropic enters our index at No. 2, a placement that reads as benchmark of the field. The composite rests on 1,105 reviewer notes gathered across the public record, and it is best understood not as a score out of ten but as a standing among peers. The trend line is pointed upward: reviewer sentiment has strengthened over recent quarters.
In its favour
Long-context reasoning, strong safety posture and enterprise trust.
On the numbers that flatter it: proprietary access, 3 named compliance attestations, and a reference throughput of ~55 tokens per second.
Points of concern
Own API dialect; no open-weight self-hosting path.
The caveat a buyer should price in before committing production traffic. As with every entry, we weigh it against the field rather than against perfection.
Incident & Uptime Note
The record of being there
Anthropic states an availability of 99.8%. Held over a full year, that headline implies on the order of 17.5 hours of cumulative unavailability — a useful sense of scale, though real incidents cluster rather than spread evenly, and a single bad afternoon can outweigh a quiet quarter.
Observatory Panel
Latency & throughput, in distribution
Monthly panels across the models Anthropic serves medium
Claude Opus
| Window | TTFT ms — mean · p50 · p90 | Throughput tok/s — mean · p50 · p90 | ||||
|---|---|---|---|---|---|---|
| May 2026 | ~1520 | ~1330 | ~2120 | ~52 | ~54 | ~63 |
| June 2026 | ~1440 | ~1290 | ~2460 | ~51 | ~55 | ~59 |
| July 2026 | ~1440 | ~1330 | ~2470 | ~51 | ~54 | ~61 |
Claude Sonnet
| Window | TTFT ms — mean · p50 · p90 | Throughput tok/s — mean · p50 · p90 | ||||
|---|---|---|---|---|---|---|
| May 2026 | ~1180 | ~1060 | ~1950 | ~65 | ~66 | ~73 |
| June 2026 | ~1240 | ~1060 | ~2010 | ~62 | ~66 | ~70 |
| July 2026 | ~1270 | ~1080 | ~2110 | ~60 | ~65 | ~75 |
Claude Haiku
| Window | TTFT ms — mean · p50 · p90 | Throughput tok/s — mean · p50 · p90 | ||||
|---|---|---|---|---|---|---|
| May 2026 | ~1220 | ~1080 | ~1970 | ~87 | ~94 | ~107 |
| June 2026 | ~1340 | ~1180 | ~2360 | ~80 | ~85 | ~98 |
| July 2026 | ~1240 | ~1060 | ~2190 | ~89 | ~96 | ~106 |
All figures are single-stream, per-request rates as one user would see them — never aggregate accelerator throughput. Serving modes are not comparable with one another: shared serverless APIs batch many tenants per accelerator, dedicated and self-served deployments hand one tenant the whole card, and specialist silicon is a regime of its own. Read each figure within its mode.
No field returns on record for Anthropic yet — the ledger above opens as soon as the first practitioner submission clears verification.
Key Facts
The record, in brief
| Category | Frontier Lab |
|---|---|
| Headquarters | San Francisco, USA |
| Founded | 2021 |
| Model access | Proprietary |
| Flagship / reference | Claude family |
| OpenAI-compatible API | No |
| Indicative price (in / out) | $3.00 / 1M tokens / $15.00 / 1M tokens medium |
| Reference throughput | ~55 tok/s |
| Reference TTFT | ~1300 ms |
| Stated uptime | 99.8% |
| Compliance | SOC 2 Type II · GDPR · HIPAA |
| Composite reputation | 90 / 100 · ≈ 4.5 / 5 · 1,105 notes |
Reviewer Notes
What the field says
The 1,105 notes behind Anthropic's standing are drawn from the accumulated public verdict of practitioners — the recurring praises and the recurring gripes — rather than from any single survey. Two themes dominate. Admirers return to one point above all: Long-context reasoning, strong safety posture and enterprise trust. Detractors return to another: Own API dialect; no open-weight self-hosting path.
We publish neither individual reviews nor reviewer identities; the composite is an editorial synthesis, and the reviewer-note count is an order-of-magnitude indication of how much public signal informs it.
Related dossiers · see also
Peer entries in the same or adjacent category, for comparison against Anthropic's standing.