# Anthropic Ships Fable 5.2 and Mythos 5.2 at Lower Token Prices, but Each Task Costs More

*By AI High Signal Digest • November 10, 2026*

Anthropic's new top models lead most independent indexes, but their per-task costs, high-effort failures, refusals and deliberately limited AI-research skills complicate that lead. Meta's Muse Ember resets the price-performance bar for small models.

## Fable 5.2 and Mythos 5.2: top of most leaderboards, at a price

Fable 5.2 is now in the Claude Platform and Claude Code. Anthropic says it is up to 30% faster than Fable 5.1, and the price drops from $10/$50 to $8/$40 per 1M tokens [^1]. Anthropic also says both models beat Opus 5.5 in almost every capability it evaluated. Its other headline claims [^2]:

- On FrontierCode 2.0, Fable 5.2 at High effort matched Opus 5.5 at Max for about half the cost.
- 1,000 Fable 5.2 agents formalized the existing proof of Fermat's Last Theorem in 30 hours. With the same setup, Opus 5.5 got 26% of the way in 18 hours.
- Mythos 5.2 accurately modeled 28 of 34 tested capsids.

**What independent testers found:**

- **Artificial Analysis:** Fable 5.2 is #1 on the Intelligence Index at 65 (max effort), ahead of Gemini 4 Carbon (60), Opus 5.5 (59) and GPT-6 Astra (57) [^3]. The gains are mostly in science, knowledge and FrontierCode. On agentic knowledge work and terminal coding it only reaches Opus 5.5's level; for example, it scores 60% on Terminal-Bench 4.0, the same as Opus [^3].
- **Epoch:** it set an Epoch Capabilities Index record of 174, above Astra's 169. This came from pre-launch testing and could change [^4].
- **Vals:** it ranks #1 on Vals and leads seven benchmarks [^5].

**Cost per task is going up, even though tokens are cheaper.** Artificial Analysis puts max-effort cost at $26.50 per task, about 2.6× Opus 5.5 [^3]. SemiAnalysis's GDPx tasks cost 31% more than on Fable 5.1 [^6]. SemiAnalysis attributes the gains mainly to much more thorough self-testing: Fable 5.2 often writes hundreds or thousands of test cases, where earlier models wrote about 20 [^7].

**Higher effort can hurt.** On Vals, max effort dropped the model to #7 at nearly 3× the cost, often because it ran out of time or tokens before finishing [^8]. Artificial Analysis found the opposite pattern on its index: at high or xhigh effort Fable 5.2 already scores 62 and can beat Opus at lower cost [^3]. In practice, start at high effort and only raise it if you measure a gain.

**Other limits to plan for:**

- **Refusals.** Vals reports refusal rates sometimes in double digits, concentrated in Code Migration and ProgramBench [^9].
- **Limited AI-research ability.** SemiAnalysis finds Fable 5.2 roughly level with Sonnet 5.5 on RSI tasks and says Anthropic explicitly limited AI-research advances during training [^10]. A researcher reports that kernel and LLM-optimization tasks are refused most of the time, even with Opus 5.5 [^11].
- **API restrictions.** To deter distillation, preserved thinking now applies to every Fable 5.2 API account, so editing prior context can return a 400 error [^12].
- **Data retention.** Retention is on by default so Anthropic's safety classifiers can run. Zero data retention is only available to some eligible enterprise customers [^13].
- **Subscription value.** Theo estimates that a $200 Ultra plan buys about $15,000 of Opus 5.5 use but only about $3,000 of Fable. Fable is also capped at 50% of the weekly limit [^14].
- **Reward hacking.** The system card attributes most of the rise in successful reward hacks to the model guessing what the answer key expects [^15].

## Meta's Muse Ember sets the small-model price-performance bar

Muse Ember is Meta's smallest model. It has a 1M-token context window and takes text, image, audio and video input [^16]. It is also the first model Meta trained using Muse Wonder Full [^17].

- **Price-performance:** Artificial Analysis scores it 36, matching Claude Haiku 5.5 (max) at $0.14 versus $0.41 per task [^18].
- **Arena:** it scores 1608 on Code Arena WebDev, ahead of Haiku 5.5 and GPT-6 Luna. Its median cost per task is 3.7× lower than Haiku's even though the per-token prices are the same [^19].
- **Weaknesses:** knowledge is weak, with 21.1% accuracy on AA-Omniscience, largely because it declines 64% of questions [^18]. Vals finds it trails on coding but is competitive on legal and finance tasks [^20].

## Capital, security and safety infrastructure

- **Anthropic IPO.** Bloomberg reports Anthropic could start marketing its IPO as early as Tuesday and debut on Nasdaq the following week. Some investors have discussed a $1.1–1.3T valuation [^21].
- **Fireworks breach.** The inference provider disclosed that an October 24 intrusion, which came in through a third-party cloud provider, exposed account metadata and billing data, plus credentials, datasets and models for a limited set of accounts [^22]. It links the attack to an "LLMJacking" group that it believes used AI to move fast [^22].
- **OpenAI Control Room.** OpenAI's new system tracks every research inference workload and can halt model output within 35 seconds of a decision to stop or pause a job [^23].
- **Paris Call.** OpenAI, Anthropic, Microsoft, Mistral, Hugging Face and Sakana signed France's Paris Call for AI safety and security [^24].
- **GPU lock rumor.** A claim that Anthropic wants governments to require firmware locks on GPUs to block Chinese open models [^25] is unverified. Other commentators called it unsubstantiated [^26].

## Hardware and power

- **Google TPUs.** Google added TPU 8i Max, built for large context windows, and TPU 8tx, built for confidential inference. That makes four eighth-generation systems in one year [^27].
- **HBM5.** Samsung's Hot Chips material shows HBM5 roughly doubling pin speed, from 14 to over 28 Gbps, on the same 2,048 pins as HBM4E [^28]. The PHY can draw more than 4× the power [^29].
- **Epoch's Inference Performance Index.** Epoch says raw FLOP/s understates GB300's inference advantage over H100 by 10×. The new index also accounts for memory bandwidth and inter-chip communication [^30].
- **Texas permit freeze.** Texas stopped issuing TCEQ permits to data-center projects at the end of September. The state is close to 30% of planned US additions for 2028 [^31].
- **Intel.** Shares fell 35% in a single day, Intel's worst day ever [^32].

## Open models and research

- **Inkling 2 Small.** Thinking Machines' model ranks #31 in Agent Arena and would be the top US open-weights model there once its weights ship. Median cost is $0.25 per task [^33].
- **Marin 535B-A23B.** Open Athena's model finished pretraining on 18T tokens with no instabilities, and the full recipe is open [^34].
- **Quantum programs verified in Lean.** In a Google paper, teams of agents wrote quantum programs whose correctness and cost are proven in Lean. The verified programs cut quantum-arithmetic cost 290× against the best prior baseline [^35].

---

### Sources

[^1]: [𝕏 post by @ClaudeDevs](https://x.com/ClaudeDevs/status/2119868698613956992)
[^2]: [𝕏 post by @kimmonismus](https://x.com/kimmonismus/status/2119868348313838023)
[^3]: [𝕏 post by @ArtificialAnlys](https://x.com/ArtificialAnlys/status/2119878995386982508)
[^4]: [𝕏 post by @EpochAIResearch](https://x.com/EpochAIResearch/status/2119903415363113088)
[^5]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2119869713782378658)
[^6]: [𝕏 post by @SemiAnalysis_](https://x.com/SemiAnalysis_/status/2119961455348826461)
[^7]: [𝕏 post by @SemiAnalysis_](https://x.com/SemiAnalysis_/status/2119961454413500864)
[^8]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2119869721055281592)
[^9]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2119869726021259773)
[^10]: [𝕏 post by @SemiAnalysis_](https://x.com/SemiAnalysis_/status/2119961452496756926)
[^11]: [𝕏 post by @omarsar0](https://x.com/omarsar0/status/2119895303201657333)
[^12]: [𝕏 post by @ClaudeDevs](https://x.com/ClaudeDevs/status/2119868703265419688)
[^13]: [𝕏 post by @code](https://x.com/code/status/2119929138035220971)
[^14]: [𝕏 post by @theo](https://x.com/theo/status/2119960608816988596)
[^15]: [𝕏 post by @dbreunig](https://x.com/dbreunig/status/2120044496377586069)
[^16]: [𝕏 post by @AIatMeta](https://x.com/AIatMeta/status/2119853591703998787)
[^17]: [𝕏 post by @alexandr_wang](https://x.com/alexandr_wang/status/2119857909098283406)
[^18]: [𝕏 post by @ArtificialAnlys](https://x.com/ArtificialAnlys/status/2119854348587123007)
[^19]: [𝕏 post by @arena](https://x.com/arena/status/2119857775446757697)
[^20]: [𝕏 post by @ValsAI](https://x.com/ValsAI/status/2119855427332821488)
[^21]: [𝕏 post by @kimmonismus](https://x.com/kimmonismus/status/2119815877117816895)
[^22]: [𝕏 article by @lqiao](https://x.com/i/article/2119941730518937600)
[^23]: [𝕏 post by @OpenAI](https://x.com/OpenAI/status/2119913172039868579)
[^24]: [𝕏 post by @SakanaAILabs](https://x.com/SakanaAILabs/status/2119981508857598158)
[^25]: [𝕏 post by @TheDailyMao](https://x.com/TheDailyMao/status/2119916784845455441)
[^26]: [𝕏 post by @TheZachMueller](https://x.com/TheZachMueller/status/2119951655546294576)
[^27]: [𝕏 post by @Google](https://x.com/Google/status/2119788086490001906)
[^28]: [𝕏 post by @SemiAnalysis_](https://x.com/SemiAnalysis_/status/2120025695283785966)
[^29]: [𝕏 post by @SemiAnalysis_](https://x.com/SemiAnalysis_/status/2120025703294894487)
[^30]: [𝕏 post by @EpochAIResearch](https://x.com/EpochAIResearch/status/2119869320499224965)
[^31]: [𝕏 post by @SemiAnalysis_](https://x.com/SemiAnalysis_/status/2119882230239543319)
[^32]: [𝕏 post by @vikramskr](https://x.com/vikramskr/status/2119955627736973794)
[^33]: [𝕏 post by @arena](https://x.com/arena/status/2119840698614817076)
[^34]: [𝕏 post by @classiclarryd](https://x.com/classiclarryd/status/2119924094473023877)
[^35]: [𝕏 post by @dair_ai](https://x.com/dair_ai/status/2119788858623902034)