Raphi-2Code retweeted
I almost forgot my favourite GPT-6.2-Astra generation - this is SVG (yes SVG!!) of a peacock - no other model came close to the precision, lifelike nature and detail of this generation, incl. Fable
Raphi-2Code retweeted
π¨GPT-6.2 Astra is now 20% cheaper so it's $8/$40 per 1m
GPT-6.2 Sol keeps the exact same price as GPT-6.1 Sol has now too
Raphi-2Code retweeted
Gemini 4 Neon and Oxygen were spotted these last two days. RSI π
@aitrackerbot
Raphi-2Code retweeted
So Qwen4 Max scores 43 on the AA intelligence index π
It's heavily distilled from Astra, like the UI's it makes are almost identical, not a very good drop imo
Google is working on a GIF feature for Gemini that animates images, following OpenAI's recent GIF launch with Images 2.6.
h/t @Liammagnier7
Raphi-2Code retweeted
Gemini 4 Oxygen (the next flash model)
Left is xhigh Hydrogen, the right is what was routing in Arena, same prompt and there is a decent difference
Raphi-2Code retweeted
It has been 72 days since OpenAI last released a frontier model - officially the longest gap since the wait between GPT-5 to 5.1 (Aug-Nov 2025)
If it doesnβt release in the next few weeks, it would become the longest gap between two OpenAI frontier model releases since o3 to GPT-5 (Apr-Aug 2025)
Raphi-2Code retweeted
Announcing Artificial Analysis Coding Agent Index v2.1, upgrading FrontierCode and Terminal-Bench. Claude Fable 5.2 and Claude Opus 5.5 share the lead at 58
The Coding Agent Index measures models in agent harnesses; v2.1 of the Index upgrades FrontierCode from 1.1 to 2.0 and Terminal-Bench from 4.0 to 4.1.
Upgraded:
β€ FrontierCode-AA 2.0 (replacing FrontierCode v1.1): our implementation of Cognition's FrontierCode 2.0, comprising 132 real-world coding tasks graded on whether a repository maintainer would merge the agent's change
β€ Terminal-Bench 4.1 (replacing Terminal-Bench 4.0): the same 66 tasks, scored with updated verifiers that improve reliability and resistance to reward hacking
Staying in the Index: SWE-Atlas-QnA-AA
Key results:
β€ Claude Fable 5.2 and Claude Opus 5.5 share the lead at 58: Claude Fable 5.2 scores 58 at both xhigh and high in Claude Code. Claude Opus 5.5 (max) scores 58 in Claude Code and, paired with GPT-6 Luna High, in Cognition's multi-model harness Devin Fusion CLI. Claude Fable 5.2 (max) scores 56 as more of its attempts hit task time limits. Claude Sonnet 5.5 (max, 56) in Claude Code and Gemini 4 Carbon (high, 55) in Antigravity CLI follow
β€ Index scores are lower in v2.1, driven by the upgrade to FrontierCode-AA 2.0: models score lower by 5 points on average. Codex with GPT-6.1 Terra (max) is the only configuration which didnβt fall between versions, sitting within 1 point of its prior score
β€ Cost per task varies ~30x among configurations scoring 48+: from $1.51 for Codex with GPT-6.1 Sol (max, 52) to $44.22 for Claude Code with Claude Fable 5.2 (max, 56)
Raphi-2Code retweeted
META π₯: Muse-Spark-1.4-internal has been spotted in one of the PRs on Github.
> The PR is a small logging change to RocksDB's experimental "live tuning" feature.
> The mention appears to be in the PR notes, which seem to be auto-generated.
It's expected that Muse Spark 1.4 is in the works and is already used internally at Meta, as it seems.
Raphi-2Code retweeted
It's crazy to think that, somehow, Anthropic has the most creative image gen model, despite having no image gen models
Raphi-2Code retweeted
Claude Fable 5.2 is live on MineBench.ai, with significant efficiency jumps over Fable 5.1!
- Total Cost (15 builds): $147.55 β $38.08 (-74%)
- Avg Cost / Build: $9.84 β $2.54
- Avg Inference Time: 40m 11s β 12m 55s (3.1x faster)
- Failed Attempts / Retries: 8 β 0 (15/15 one-shotted)
The massive reduction in cost is a result of Fable 5.2's lower API pricing, but also the model requiring zero retries in our harness (15 total generations vs. Fable 5.1's 23).
Raphi-2Code retweeted
Gemini 4 Carbon is now on MineBench.ai :)
- Total Cost: $11.65
- Average Inference Time: 16m 0.9s
Gemini 4 Argon (for reference):
- Total Cost: $9.12
- Average Inference Time: 6m 15.1s
Though its ratings are yet to stablize, it currently overtakes Argon for 2nd place on our leaderboards!
Raphi-2Code retweeted
Looks like it's between Terra and Sonnet until you turn on subsidization
Google's new Gemini 4 Hydrogen has Intelligence just below Claude Sonnet 5.5 and GPT-6.1 Sol, at a third of Sonnetβs cost but ~2.5x that of GPT-6.1 Sol
@GoogleDeepMind has released their third new model in six weeks. Gemini 4 Hydrogen introduces an xhigh reasoning level and scores 54 on the Artificial Analysis Intelligence Index, with strengths in visual reasoning and non-hallucination. This places it 1 point behind Claude Sonnet 5.5 (max) and GPT-6.1 Sol (max), at ~32% of Sonnetβs cost but ~2.5x that of GPT-6.1 Sol. With high reasoning, Gemini 4 Hydrogen matches the Intelligence of Gemini 4 Argon for ~19% of the Cost per Task
Key takeaways:
β€ The new xhigh reasoning level scores 54 on the Intelligence Index: xhigh is the first reasoning level above high on a Gemini model, and puts Hydrogen 1 point below Claude Sonnet 5.5 (max) and GPT-6.1 Sol (max), and 4 points ahead of Gemini 4 Argon (high). With default (high) reasoning, Gemini 4 Hydrogen scores 51, just behind GPT-6.1 Sol (xhigh, 52) and Claude Opus 5.5 (high, 52)
β€ Strong visual reasoning, presentation and non-hallucination: On GDP.pdf, a visual reasoning benchmark over long documents, it scores 33% all-pass on medium reasoning, the tied #1 score on the benchmark. On AA-Omniscience, Hydrogen has one of the lowest hallucination rates recorded at 8%, compared with 43% for Claude Sonnet 5.5 (max) and 52% for GPT-6.1 Sol (max). On AA-Briefcase, our agentic knowledge work evaluation, graders rate Hydrogen's presentation above other Gemini models, including Carbon
β€ Weaker on agentic terminal work: Hydrogen scores 54% on Terminal-Bench 4.0 and 50% on Terminal-Bench-Science 0.1, covering agentic coding and scientific research workflows in the terminal. Claude Sonnet 5.5 (max) scores 64% and 53%, and GPT-6.1 Sol (max) scores 56% and 58%. xhigh reasoning helps on the science tasks, lifting Terminal-Bench-Science from 41% at high to 50%, but makes no difference on Terminal-Bench 4.0 (54% at both). Science scores are even across domains, from 46% in earth sciences to 53% in physical sciences
β€ Low cost for its intelligence, with high token use at xhigh: Hydrogen is priced at $1.50/$7.50 per million input/output tokens. At xhigh it costs ~32% as much per task as Claude Sonnet 5.5 (max) for near-equal intelligence, but ~2.5x as much as GPT-6.1 Sol (max). xhigh nearly doubles output tokens compared with high for a 3 point gain in Intelligence, and Cost per Task falls from $3.16 at xhigh to $1.87 at high. Medium ($0.80) and low ($0.29) cost less again, but score 41 and 27 on the Intelligence Index
β€ On the Time per Task vs. Intelligence Pareto frontier: Gemini 4 Hydrogen has an output speed of ~200 output tokens per second. With high, medium, and low reasoning effort, Gemini 4 Hydrogen is on the Time per Task vs. Intelligence Pareto frontier, meaning no models are both more intelligent and have a lower Time per Task
Key model details:
β€ Context Window: 1M tokens
β€ Multimodality: Text, image, video, and speech input, with text output
β€ Reasoning efforts: Gemini 4 Hydrogen offers four reasoning levels (low, medium, high, and xhigh) with xhigh new in this release
β€ Pricing: $1.50/$7.50 per 1M input/output tokens. Cached input tokens receive a 95% discount ($0.075)
Congratulations to the @GoogleDeepMind team on the launch!
Raphi-2Code retweeted
Gemini 4 Hydrogen (xHigh) scores 54 on the AA-Intelligence Index, 4 points higher than Gemini 4 Argon and 1 point behind Sol 6.1
























































