Anthropic's 2026-06 frontier model — top-tier agentic coding and reasoning, 1M context, 128K output.
Claude Fable 5 is Anthropic's June 2026 frontier Claude model for the heaviest reasoning, coding, agentic, and long-context workloads. In TheRouter it is cataloged as anthropic/claude-fable-5, with text and image input, text output, a 1M-token context window, and up to 128K output tokens flowing from the live model catalog.
The operational story matters as much as the headline capability. Anthropic's launch introduced a new model id, always-on adaptive thinking, refusal-aware server-side fallback semantics, and deployment across Claude API, AWS Bedrock, Vertex AI, Microsoft Foundry, and later Snowflake Cortex. Subsequent nationality controls and credit-pricing changes mean routers should treat Fable 5 as a premium route with explicit fallback, billing, and eligibility policy rather than as a silent default.
Anthropic operates Fable 5 natively through Claude API semantics, including adaptive thinking, refusal handling, and provider-specific deployments such as Bedrock, Vertex AI, Microsoft Foundry, and Snowflake Cortex.
TheRouter serves it as the catalog model id anthropic/claude-fable-5 on the OpenAI-compatible chat completions path; operational pricing, limits, and supported parameters are taken from the live TheRouter catalog, not copied into this page.
| Type | Rate |
|---|---|
| Input | $10.80 / 1M tokens |
| Output | $54.00 / 1M tokens |
| Release date | 2026-06-09platform.claude.com ↗ | verified |
| Anthropic model id | claude-fable-5platform.claude.com ↗ | verified |
| Context window | 1,000,000 tokens | verified |
| Maximum output | 128,000 tokens | verified |
| Training cutoff | Not publicly disclosed | unknown |
| Deployment | Claude API, AWS Bedrock, Google Cloud Vertex AI, Microsoft Foundry, Snowflake Cortexwww.snowflake.com ↗ | verified |
| License | Anthropic Usage Policy (proprietary, API-only)www.anthropic.com ↗ | verified |
| Benchmark | Distribution | Score | Source |
|---|---|---|---|
Official benchmark table Anthropic's launch documentation describes Fable 5 as its strongest public Claude route but does not publish a benchmark table in the cited model documentation. | — | Not publicly disclosed | — |
TheRouter live snippet verification No snippet execution ledger exists for this page yet; the scheduled curator is not authorized to spend API budget on verification. | — | Not publicly disclosed | — |
Independent benchmark table No authoritative independent score is recorded in this curation pass; unknown is retained rather than copying unverified aggregator claims. | — | Not publicly disclosed | — |
Use the global api.therouter.ai endpoint shown below for new integrations; the legacy China accelerated endpoint is retired.
curl https://api.therouter.ai/v1/chat/completions -H "Content-Type: application/json" -H "Authorization: Bearer $THE_ROUTER_API_KEY" -d '{
"model": "anthropic/claude-fable-5",
"messages": [
{"role": "user", "content": "Summarize the key points from this input."}
]
}'Use the OpenAI-compatible chat completions endpoint when you want TheRouter to route Claude Fable 5 through its configured provider chain.
curl https://api.therouter.ai/v1/chat/completions \
-H "Authorization: Bearer $THEROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"anthropic/claude-fable-5","messages":[{"role":"user","content":"Hello"}]}'TheRouter's June 22 operator note explains why restored access still requires explicit eligibility checks, fallback policy, and billing alerts before routing production traffic to Fable 5.
No. It is a premium route for high-stakes reasoning, coding, agentic, and long-context tasks. For ordinary chat or high-volume automation, route to Haiku, Sonnet, or a cheaper cross-provider model first.
The scheduled curator has no API budget and did not run snippet verification. Only a human-approved verification ledger entry can raise that 20% component.
At minimum, log requested model id, returned model id, provider route, refusal/fallback outcome when available, token usage, and customer billing tier. The model's June 2026 rollout showed that silent routing or pricing assumptions can break quickly.
| source | URL | retrieved | |
|---|---|---|---|
| Release date | platform.claude.com ↗ | 2026-07-30 | verified |
| Anthropic model id | platform.claude.com ↗ | 2026-07-30 | verified |
| Context window | — | — | verified |
| Maximum output | — | — | verified |
| Training cutoff | — | — | unknown |
| Deployment | www.snowflake.com ↗ | 2026-07-30 | verified |
| License | www.anthropic.com ↗ | 2026-07-30 | verified |
| Official benchmark table | platform.claude.com ↗ | 2026-07-30 | unknown |
| TheRouter live snippet verification | therouter.ai ↗ | 2026-07-30 | unknown |
| Independent benchmark table | www.anthropic.com ↗ | 2026-07-30 | unknown |
| Claude Fable 5 returns with nationality controls and a credit pricing cliff | therouter.ai/news ↗ | 2026-07-30 | verified |
| Is Claude Fable 5 a good default model? | platform.claude.com ↗ | 2026-07-30 | to verify |
| What should a router log for Fable 5 traffic? | therouter.ai ↗ | 2026-07-30 | to verify |