Long-Horizon Autonomy
Tackle projects that take days with a proactive model that plans across stages and delegates to sub-agents.
Claude Fable 5
The cheap Claude-Fable-5 API offers Mythos-level intelligence for your hardest projects. It excels at autonomous coding, multi-day reasoning, and complex vision tasks, all while maintaining high safety standards and token efficiency.
Voorbeelden van API-gebruik
curl --request POST "https://gptproto.com/v1/chat/completions" \
--header "Authorization: Bearer $GPTPROTO_API_KEY" \
--header "Content-Type: application/json" \
--data '{
"model": "claude-fable-5",
"messages": [
{
"role": "user",
"content": "Hello"
}
]
}'Claude Fable 5 prijzen
Chatten, codeeragents en documentwerk. Geprijsd per 1 miljoen tokens — invoer, gecachte invoer en uitvoer worden afzonderlijk berekend. GPTProto is 10% goedkoper dan de officiële tarieven.
| Scenario | Claude prijslijst | OpenRouter | GPTProto | Je bespaart / mnd |
|---|---|---|---|---|
| Persoonlijk10M tokens / mnd (4.8M gecachet) | $136.80 | $144.32 | $123.12 | −$13.68≈ $164.16 / jr |
| Team100M tokens / mnd (48M gecachet) | $1368.00 | $1443.24 | $1231.20 | −$136.80≈ $1641.60 / jr |
| Zakelijk500M tokens / mnd (240M gecachet) | $6840.00 | $7216.20 | $6156.00 | −$684.00≈ $8208.00 / jr |
Bespaar $0.00 / maand
Jaarlijkse besparing ≈ $0.00 · betalen naar gebruik
Gerelateerde modellen
Alle modellen
Claude Fable 5Huidig
Claude · 1M context · $9/M input · $45/M output
Bytedance · 262K context · $0.0971/M input · $0.9714/M output
Bytedance · 262K context · $0.0971/M input · $0.9714/M output
Model
Prijs: GPTProto
vs. officieel
vs. OpenRouter
Context
Stabiliteit
Claude Fable 5Huidig
Claude
$9.00 / $45.00per 1 miljoen tokens
-10%
-15%
1M
Fable 5: Mythos Power for Everyone
Why a cheap Claude-Fable-5 API is the ultimate tool for frontier intelligence.
Senior-Grade Reasoning
Achieve senior research scientist grade outputs, picking directions and producing novel first-principles results.
Token Efficiency
Handle 50-million-line codebases with fewer turns and higher reliability than any previous Claude iteration.
Vision-First Audit
Extract precise numbers from scientific figures or rebuild source code from screenshots with zero scaffolding.
What is Claude Fable 5?
Claude Fable 5 is the first model in Anthropic's Claude 5 family and its first Mythos-class model released to the public, a capability tier above Claude Opus. It's state-of-the-art on most published benchmarks for software engineering, knowledge work, vision, and scientific reasoning, and its lead grows the longer and more multi-step the task. Launched June 9, 2026.
|
Spec |
Value |
|
Model string |
claude-fable-5 |
|
Context window |
1,000,000 tokens |
|
Max output |
128,000 tokens / request |
|
Input |
Text + image + file |
|
Output |
Text |
|
Reasoning |
Summarized thinking (raw chain-of-thought not returned) |
|
Provider |
Anthropic |
|
GPTProto endpoint |
https://gptproto.com/v1/messages |
Claude Fable 5 vs Opus 4.8 vs GPT-5.5
|
|
Claude Fable 5 |
Claude Opus 4.8 |
GPT-5.5 |
|
SWE-Bench Pro |
80.3% |
69.2% |
58.6% |
|
Context window |
1M |
1M |
1M |
|
Input |
Text + image + file |
Text + image |
Text + image |
|
Best for |
Long-horizon agentic work, migrations, vision audits |
General reasoning; cheapest of the three |
Codex-style coding workflows |
|
GPTProto price (in / out per 1M) |
$8 / $40 |
$4 / $20 |
$4 / $24 |
Benchmarks as reported by Anthropic / OpenAI; validate on your own workload before committing. The honest read: Fable 5 tops the benchmarks but runs ~2× the price of the others. Opus 4.8 is the cheapest of the three and the better default for most general reasoning. GPT-5.5 sits in the middle on price and is the pick if you're standardized on OpenAI / Codex tooling. All three run from one GPTProto key and balance, so you can route each task to the right model without switching providers.
Claude-Fable-5 FAQ: Cheap API & Safeguards
Answers about the cheap Claude-Fable-5 API, including pricing models, safeguard fallbacks, and autonomous agent capabilities for your enterprise.
How much does the cheap Claude-Fable-5 API cost?
Claude Fable 5 is priced at $10 per million input tokens and $50 per million output tokens. This is less than half the price of the previous Mythos Preview. Additionally, users can leverage a 90% discount on input tokens through prompt caching, making it a very cheap Claude-Fable-5 API option for high-volume, repetitive tasks that require persistent context or large document sets stored in memory.
When should I use Fable 5 instead of Opus 4.8?
Use Fable 5 for your most ambitious, long-running projects that require autonomy. While Opus 4.8 is excellent for general reasoning, Fable 5 is built for tasks that span days, such as large-scale code migrations or deep research. If your query is flagged by cybersecurity or biology safeguards, the system will automatically route to Opus 4.8 to ensure safety without interrupting your workflow or charging you Fable prices.
Does Fable 5 support autonomous coding?
Yes, it is Anthropic's most capable model for agentic coding. It can perform codebase-wide migrations, write its own tests, and use vision to compare outputs against design goals. Stripe reported that Fable 5 compressed months of manual engineering work into just a few days. It functions effectively in agent harnesses like Claude Code, planning across multiple stages and delegating to sub-agents without supervision.
How do the vision capabilities compare?
Fable 5 is the state-of-the-art for vision tasks. It extracts precise data from complex charts, diagrams, and tables within PDFs. Unlike earlier models that required complex harnesses to interpret game states, Fable 5 has demonstrated the ability to beat games like Pokémon FireRed using only raw screenshots. This makes it ideal for rebuilding web apps from UI captures or auditing document-heavy files in legal sectors.
What are the data retention policies?
To maintain high safety standards for Mythos-class models, Fable 5 requires a 30-day data retention period for safety monitoring. This applies to both first-party and third-party surfaces. This data is not used for training new models; instead, it helps identify and reduce false positives in the safeguard classifiers. All human access to this logged data is strictly monitored and recorded to protect user privacy.
Is US-only inference available?
For organizations with strict data residency requirements, US-only inference is available. This ensures that the model processing stays within US borders. This specialized routing is priced at 1.1x the standard rate for both input and output tokens. This means you can still access a relatively cheap Claude-Fable-5 API while satisfying the compliance needs of legal, financial, or government-contracted projects.
Gerelateerde artikelen
Gidsen, vergelijkingen en updates over dit model.
Alle artikelen