New Sign up free, 10 calls on us. Up to $1, no card needed.

Claude Fable 5

Released 2026-06-09

invited beta chatCodeReasoningTool callingVisionPrompt caching

Claude Fable 5 is Anthropic's most capable widely released model, built for the most demanding reasoning and long-horizon agentic work, described as "next-generation intelligence for long-running agents."

Input
text image $10/M
Output
text $50/M
Cache read
$1/M
Context
200K
Knowledge cutoff
2026-01

Benchmarks

Above averageNo peer higher74 / 7941 / 79
Claude Fable 5 other models measured peer average no peer scored higher
SWE-Bench Pro
no peer scored higher 80.3%
BioMysteryBench hard
46.5%
OSWorld-Verified
no peer scored higher 85%
Cybergym
83.1%
HealthBench Professional
no peer scored higher 60.9%
GDPval-AA v2 Elo · 642-1861
1747
Harvey Lab-AA
93.6%
GPQA Diamond
92.6%
Blueprint-Bench 2
38.6%
BrowseComp
87.4%
CharXiv (RQ) reasoning
no peer scored higher 88.9%

Vendor-published: Alibaba (Qwen) Anthropic ByteDance DeepSeek Google MiniMax Moonshot OpenAI Tencent Z.ai

Price in context

Where the price sits among 60 comparable models

Input$10/M
$0.05 · Qwen3 VL Flash GPT-5.4 Pro · $30
Output$50/M
$0.275 · DeepSeek V4 Flash GPT-5.4 Pro · $180
Cached read$1/M
$0.0028 · DeepSeek V4 Flash GPT-5.4 Pro · $15

The bar shows how this model’s price compares with every other model of the same kind on Synthorai. The cheapest and the most expensive are named at each end. These are base rates; batch, region and cache-write discounts are on the pricing page.

Specs & limits

Tokens

Context window (vendor spec) 1,000,000
Max output (vendor spec) 128,000
Knowledge cutoff 2026-01

Prompt caching

How it caches explicit (opt-in)
Min prefix 512 provider default is 1,024
Lifetime 5m default, 1h option
Write cost 1.25x (5m) / 2x (1h)

Thinking

Vendor control output_config.effort (thinking.type is adaptive-only and needs no configuration)
Accepted values effort low · medium · high · xhigh · max; both "enabled" and "disabled" return 400
Default thinking always on (adaptive); effort high applied when the request sets nothing
Can be turned off No
Thinking behaviour Thinking cannot be turned off, so use display "omitted" (the default here) to keep thinking text out of responses instead; the raw chain of thought is never returned, and a request that tries to elicit internal reasoning as response text can be refused with stop_details.category "reasoning_extraction".
Parameter reasoning_effort
Values minimal · low · medium · high the gateway's parameter surface - the vendor mapping above applies

Model

Modalities text + image → text
  • Runs adaptive thinking always-on with effort-based depth control
  • includes safety classifiers with documented fallback and billing-credit mechanics

per Anthropic official docs ↗

One prompt, measured through the gateway

PROMPT Explain to a CFO, in exactly three sentences, why our LLM bill doubled after we enabled reasoning. CHECK exactly 3 sentences

Claude Fable 5 passed · 3 sentences

When we enabled reasoning, the model began generating extensive internal "thinking" tokens before each answer, and those tokens are billed as output even though users never see them. Output tokens typically cost several times more than input tokens, so adding thousands of hidden reasoning tokens per request effectively multiplied our per-query cost. In short, we're paying for a much longer response behind the scenes in exchange for higher answer quality, and we can manage this by tuning reasoning effort levels or reserving reasoning mode for complex tasks only.

out 228 tok latency 6.7 s

Instruction following (exactly three sentences - countable), audience shaping (CFO register), and the hidden-thinking billing gap the token meters below expose.

PROMPT This function is supposed to return the longest run of consecutive equal items, but callers report it is off by one on single-element inputs and misses runs that end at the last index. Fix it and explain the bug in one paragraph. CHECK fix passes tests

Claude Fable 5 passed · 8/8 cases

## Fixed code ```python def longest_run(xs): if not xs: return 0 best = cur = 1 for i in range(1, len(xs)): if xs[i] == xs[i - 1]: cur += 1 else: cur = 1 best = max(best, cur) return best ``` ## Explanation The original function counts **adjacent equal pairs** rather than **items in a run**: a run of length *n* contains only *n − 1* equal neighbor pairs, so every result is off by one — a single-element list yields 0 instead of 1, and `[1, 2, 2]` yields 1 instead of 2 (which callers perceive as "missing" the run at the end). The fix is to initialize both `cur` and `best` to 1 (any nonempty list has a run of at least length 1), reset `cur` to 1 rather than 0 when

out 531 tok latency 12.5 s

Whether the fix is actually correct (runnable), explanation density, and token efficiency on a bounded task.

PROMPT Extract a JSON object with fields {company, quarter, revenue_usd_m, growth_pct, guidance} from this text. Use null for anything not stated; add no other fields. CHECK valid JSON, schema exact

Claude Fable 5 passed · 5/5 fields, guidance null

```json { "company": "Meridian Robotics", "quarter": "Q3 (September quarter)", "revenue_usd_m": 412, "growth_pct": 11, "guidance": null } ``` Notes on interpretation: "up eleven and change" is rendered as 11 (the precise figure isn't stated), and guidance is null because it was explicitly withheld.

out 192 tok latency 6.5 s

Schema obedience (no invented fields), hallucination pressure (guidance is explicitly withheld), and structured-output path differences.

PROMPT Write a 120-word product announcement for an API gateway feature that caches prompts across providers. Forbidden words: "seamless", "unlock", "game-changer", "revolutionize", "empower". Exactly one sentence must be a question. CHECK 120 words, 0 banned words

Claude Fable 5 passed · 120 words, 0 banned, 1 question

**Introducing Cross-Provider Prompt Caching for our API Gateway** Starting today, your repeated prompts are cached once and reused across OpenAI, Anthropic, Google, and any other provider you route through the gateway. Instead of paying full token costs every time your application sends the same system prompt or context block, the gateway detects matching prefixes and serves cached results in milliseconds. Why keep spending money and latency on identical requests? Early adopters report cost reductions of up to 60 percent on high-volume workloads, along with faster median response times. Configuration is simple: enable caching in your dashboard, set a TTL, and choose which routes participate. Cache entries are encrypted at rest and never shared between accounts. Available now on all plans.

out 1173 tok latency 18.2 s

Constraint obedience (word budget, banned-word list, the single question), style fingerprint, and length control.

Use Claude Fable 5 in 30 seconds

OpenAI-compatible: swap the base_url, keep your SDK. POST /v1/chat/completions

from openai import OpenAI

client = OpenAI(
    base_url="https://synthorai.io/v1",
    api_key="sk-syn-...",
)

resp = client.chat.completions.create(
    model="claude-fable-5",
    messages=[{"role": "user", "content": "Summarize this diff"}],
    reasoning_effort="medium",
)
print(resp.choices[0].message.content)

About Claude Fable 5

  • Anthropic's model-selection guidance reserves it for workloads that need the highest available capability, namely long-running agents, deep reasoning, and advanced research, while pointing the Opus tier at everyday complex agentic coding and enterprise work.
  • It runs adaptive thinking always-on with effort-based depth control, supports a 1M-token context window and 128K output tokens, and launches with the memory tool, code execution, programmatic tool calling, compaction, and vision.
  • Thinking cannot be switched off: both the enabled and disabled thinking types return a 400, raw chain of thought is never returned, and summaries are opt-in through the thinking display setting.
  • Effort runs from low through max including xhigh, defaults to high, and Anthropic notes that its lower effort settings often exceed the xhigh performance of earlier models.
  • Prompt caching starts at a 512-token minimum prefix, the lowest in the lineup.
  • Uniquely, it includes safety classifiers that can decline requests, with documented fallback and billing-credit mechanics.
  • Refusals arrive as a successful response carrying a refusal stop reason and a category, and a request refused before any output is generated is not billed.
  • Assistant prefill and non-default sampling parameters both return errors.
  • Synthorai serves Claude Fable 5 (currently an invite-only beta on the platform) through its OpenAI-compatible endpoint.

FAQ

Is the Claude Fable 5 API free to try?

Claude Fable 5 is currently in invited beta: access is application-based rather than open signup. Apply from the Synthorai console; once approved, standard pay-as-you-go pricing applies with no subscription.

What is Claude Fable 5 best at?

Adaptive thinking always-on with effort-based control; 1M-token context window and 128K output; safety classifiers that can decline requests. See the About section for the full picture from the vendor's own release notes.

How much does Claude Fable 5 cost?

Claude Fable 5 costs $10 per million input tokens and $50 per million output tokens on Synthorai. That is the provider's list price, with no platform markup. Cached input tokens bill at $1/M.

Does Claude Fable 5 support prompt caching?

Yes, via opt-in: mark stable prefixes with cache_control breakpoints. Cached input tokens bill at $1/M vs $10/M uncached; prompts need a 512-token stable prefix to cache (TTL 5m default, 1h option). Claude Fable 5 caching guide →

How do I get access to Claude Fable 5?

Claude Fable 5 is in invited beta: request access from the Synthorai console. Once approved it works like every other model: point your OpenAI SDK at base_url="https://synthorai.io/v1" and set model="claude-fable-5".

What is Claude Fable 5's knowledge cutoff?

Claude Fable 5's knowledge cutoff is 2026-01, per the vendor's official documentation (as of 2026-07-09).

Related models

Compare

Every value on this page is transcribed from the vendor's own documentation, linked above, and carries the date it was checked. Prices are compared across the catalogue; specification values that vendors define differently are shown with the difference stated rather than charted. Nothing here is measured by us, and nothing is scored.

Get your API key Compare your cost →