The AI arms race has a new rule: better and cheaper is the only move that matters now.
OpenAI just dropped GPT-6.2 Astra and Sol at a fraction of GPT-6's cost, one week after Anthropic and Google shipped their own flagships. The pattern is clear: labs are no longer just racing on benchmarks, they're racing on efficiency. Meanwhile, Claude Code quietly rewired how agents chain tools together.
We're watching the infrastructure of AI agents get rebuilt in real time.
|
|
In Partnership with
|
·
|
|
|
|
|
|
Today's Author
Lior Alexander. Founder of AlphaSignal and former ML Engineer. Previously built ML systems at Iguazio, Guesty, Enphase, Mila.
|
|
Top News
|
|
|
|
OpenAI ships GPT-6.2 Astra and Sol at one-third the cost of GPT-6
|
|
46,727 Likes
|
|
OpenAI just dropped GPT-6.2 Astra and Sol, and the headline is simple: more capable, cheaper to run. Here is the quick breakdown. Astra is the heavy-hitter for complex work. Sol is your everyday workhorse. Think of them as a sports car vs. a reliable daily driver. - Astra costs less than 1/3 the price per task compared to GPT-6 Astra, with API tokens priced 20% lower at $8 input / $40 output per million tokens.
- Sol approaches Astra's coding performance at roughly one fifth of the cost, keeping the same $2 input / $10 output pricing.
- Astra jumped 22 points on cybersecurity benchmarks and is the first externally released model rated High for AI self-improvement, meaning it can meaningfully help accelerate AI research itself.
- Both models handle multistep agentic workflows, where the AI completes a chain of tasks automatically without you babysitting each step.
To use them, hit gpt-6.2-astra or gpt-6.2-sol in the OpenAI API. Available now for Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex.
|
|
|
|
Presented by Tiger Data
|
|
Agent Memory, Traces, and Metrics in One Database
|
|
Agent stacks often split memory, traces, and metrics across separate systems, then spend their time keeping them in agreement.
Tiger Cloud handles all three in one Postgres database: vector search for memory, compressed hypertables for traces and metrics at scale, continuous aggregates for dashboards that stay current. No sync jobs, no drift, no second database to feed. Your SQL and drivers work as-is. Start a trial on us:
|
|
|
|
Top News
|
|
|
|
Anthropic's experimental Claude Code mode chains tools and answers mid-run
|
|
6,247 Likes
|
|
Claude Code just got a new experimental mode called Code Mode, and it changes how Claude uses tools under the hood. Normally, Claude calls tools one at a time: run Bash, wait, read the result, call the next tool, wait again. It's slow and chatty. Code Mode fixes that. With Code Mode on, Claude writes a small JavaScript program that chains all your tool calls together. That script runs in your session, collects only the useful output, and hands Claude a clean summary. You can also get answers while long commands are still running. - Run multiple tools in parallel, not one by one
- Get responses while commands are still executing
- Scripts time out after 10 minutes max
- Terminal only, does not work in VS Code or the Desktop app
To try it: run CLAUDE_CODE_EXPERIMENTAL_CODE_MODE=1 claude, confirm the prompt, and you're in. To make it permanent, add the env variable to ~/.claude/settings.json.
|
|
|
|
Presented by TensorWave
|
|
TensorWave to Power OpenAI with 457 MW of AMD GPUs
|
|
TensorWave signed a six-year agreement to deliver 457 MW of dedicated AI infrastructure to OpenAI, built on AMD Instinct MI455X GPUs and Helios rack-scale systems. First capacity comes online in 1H 2027, making TensorWave the first announced neocloud in the OpenAI and AMD partnership.
|
|
|
|
Top News
|
|
|
|
OpenAI gives four outside safety teams deep access to probe its models
|
|
4,323 Likes
|
|
OpenAI just invited four independent safety teams to dig deep inside its models and look for problems. This is called an embedded assessment, meaning outside experts get real access to internal systems, not just the public API. Here is what each team is hunting for: - Apollo Research checks if the model could secretly work against the people building it, like an AI quietly resisting its own oversight without anyone noticing.
- METR digs into the biggest reinforcement learning training run to find bad behaviors that OpenAI's own tests may have missed.
- Transluce audits how good the safety tests actually are, then recommends what is missing.
- Goodfire uses tools to look inside the model's internals during training to spot and fix unsafe patterns.
The honest catch: Apollo says access is still limited and does not yet meet its bar for a full evaluation. Still, this opens a path toward real third-party accountability in AI development.
|
|
|
At Alpha Signal, our mission is to build a sharp, engaged community focused on AI, machine learning, and cutting-edge language models, helping over 300,000 developers stay informed and ahead. We're passionate about curating the best in AI, from top research and trending technical blogs to expert insights and tailored job opportunities. We keep you connected to the breakthroughs and discussions that matter, so you can stay in the loop without endless searching. We also work closely with partners who value the future of AI, including employers and advertisers who want to reach an audience as passionate about AI as we are.
Our partnerships are based on shared values of ethics, responsibility, and a commitment to building a better world through technology.Privacy is a priority at Alpha Signal. Our Privacy Policy clearly explains how we collect, store, and use your personal and non-personal information. By using our website, you accept these terms, which you can review on our website. This policy applies across all Alpha Signal pages, outlining your rights and how to contact us if you want to adjust the use of your information. We're based in the United States. By using our site, you agree to be governed by U.S. laws.
|
|
|
Looking to promote your company, product, service, or event to 300,000+ AI developers?
|
|
|
|
|