Ahmad Raza Khan

Applied AI @AdaptiveML

[Ahmad Raza Khan @kbash09]

@egrefen too tactical to understand the head of strategic futures

[Ahmad Raza Khan @kbash09]

Wow- open source gap seems to be closing

[Arena.ai @arena]

Big news: Kimi-K3 by @Kimi_Moonshot is now #1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5.

This is a 17-place jump from Kimi-k2.6 (#18 -> #1).

In Frontend, Kimi-K3 ranked #1 in 6 of 7 domains: Brand & Marketing, Reference-Based Design, Data & Analytics, Consumer Product, Simulations, and Content Creation Tools, landing #2 only in Gaming behind Fable 5.

The full model weights will be released by July 27.

Congrats to the @Kimi_Moonshot team on this major milestone!

[Ahmad Raza Khan @kbash09]

“If you stop to think about it, we’ve essentially found a way to make sand think. It’s miraculous.”

[Demis Hassabis @demishassabis]

https://t.co/PTeDiv1b6L

[Ahmad Raza Khan @kbash09]

Really cool 🧠

[AI at Meta @AIatMeta]

We’re sharing the next major milestone in our non-invasive brain-to-text decoder research: Brain2Qwerty v2.

Building on v1, which was published today in @Nature, Brain2Qwerty v2 is the highest-performing end-to-end pipeline capable of real-time sentence decoding from raw brain signals. It advances beyond character-level performance to decoding words and semantics, enabling accuracy for overall communication.

We believe this research has the potential to make a real difference for the millions of people who suffer from brain lesions or disorders that prevent them from communicating.

🧵👇

[Ahmad Raza Khan @kbash09]

Excellent list of best practices

[Brian Armstrong @brian_armstrong]

How to keep AI spend flat while token usage grows exponentially: Not with friction and spend alerts. With better defaults, routing, and caching.

Better Defaults (not Usage Caps) – Engineers can choose any model they want, but defaults matter. We’re experimenting with defaulting to open weight models like GLM 5.2 and Kimi 2.7 through our LLM gateway, while still encouraging engineers to choose the right model for the task. 91% of our employees were never hitting their usage caps, so instead of lowering caps and driving up alerts, we're moving to cheaper defaults. Note that code reviews use a diversity of models, so they can check each other's work.

Better Routing – In our custom harnesses, we preprocess prompts and route to the best model for the job, considering cache hits and model pricing. For instance, you may want a frontier model for planning, but not for execution where they can be overkill. Ultimately, humans shouldn't be choosing models - AI can automate this task.

Better Caching – Cache misses are the easiest way to drive your cost up. All of our requests are cache aware, so we’re reusing a warm cache wherever possible. For example, our cache hit rate went from 5% → 60% in LibreChat once properly implemented.

Keep Context Lean – Start fresh sessions when switching tasks. Scope file context narrowly. Disconnect unused tools. Don't just compact. The goal isn't fewer tokens used, it's fewer tokens wasted.

Better Visibility – Our engineers can use as many tokens as they want, from whatever model they want, but we’ve made usage visible – and the more you spend on AI, the more impact we expect.

The goal isn't to suppress usage. It's to build the infrastructure that makes exponential growth sustainable.

Putting this into practice has cut our AI spend nearly in half, while our token usage continues to grow.

[Ahmad Raza Khan @kbash09]

Lol

[Ahmad Raza Khan @kbash09]

Wow, this is crazy if true

[Alexander Knigge @AlexanderKnigge]

oh my god its happening

@MistralAI has officially confirmed the upcoming release of Le Chaton Fat

- 30T MoE with 256 experts
- 1M context window
- multimodal and multilingual
- outperforms Fable 5 on every benchmark

[Ahmad Raza Khan @kbash09]

“Private reinforcement learning environments should let models grow stronger on real traces from inside the organization”

[Satya Nadella @satyanadella]

https://t.co/vLmiBKTtX3

[Ahmad Raza Khan @kbash09]

@_The_Prophet__ Every new AI restriction strengthens the case for sovereign AI and open-source models.

If access to your AI capabilities depends on someone else’s permission, it’s not truly yours.

Ownership, control, and independence will matter more than ever.

[Ahmad Raza Khan @kbash09]

Crazy

[Anthropic @AnthropicAI]

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.

Access to all other Claude models is not affected.

We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.

Read our full statement: https://t.co/bwn0sximKZ

[Ahmad Raza Khan @kbash09]

@julien_c And not even fully open but only don’t block folks from using their models for building other AI systems

[Ahmad Raza Khan @kbash09]

What an amazing talk. Must watch for all the parents and mentors out there

[Bill Gurley @bgurley]

Happy to share that my Ted Talk from Vencouver in April just went live. In the heat of job anxiety, I believe there is only one thing that makes a job “truly safe.”

https://t.co/OcpYPXH9s6

[Ahmad Raza Khan @kbash09]

Most enterprises will need to use RL to tune small language models (SLMs) to keep costs low while maintaining frontier accuracy. If you are in AI, probably best to learn post training techniques

[clem 🤗 @ClementDelangue]

Routing and post-training open-source models won't only give you more accurate systems but also meaningfully faster and cheaper systems as most companies are currently learning (in addition to giving you more control and privacy).

The idea that a "frontier" model (by frontier we mean is slightly more accurate on a few very limited benchmarks) will be better for all domains, all tasks, all setups just doesn't hold up! It's marketing for making you pay more!

[Ahmad Raza Khan @kbash09]

@Sirupsen Congrats- amazing growth!

[Ahmad Raza Khan @kbash09]

Impressive results for Gemini 3.5 Flash

[ARC Prize @arcprize]

Gemini 3.5 Flash ARC-AGI (Verified)

ARC-AGI-2:
- High: 72.1%, $0.85
- Minimal: 8.9%, $0.11

ARC-AGI-1:
- High: 92.5%, $0.42
- Minimal: 48.8%, $0.06

Gemini 3.5 Flash is on par with GPT-5.5 (Medium) on ARC-AGI

[Ahmad Raza Khan @kbash09]

We RL fine-tuned Llama 3.2 3B on Adaptive Engine to match proprietary-model function-calling accuracy.
Deployed on a single H100: 160ms server-side / 230ms client-side — 90%+ latency reduction vs. the proprietary baseline.
https://t.co/8S37VxIOer

[Adaptive ML @AdaptiveML]

In patient support, every second matters.

With @CCSMedical, we fine-tuned a 3B @metaai Llama model (on @AWSstartups) on Adaptive Engine to match proprietary model accuracy on function calling, with over 90% lower latency.

Smaller, faster, and tuned for the task.

Read the case study:
https://t.co/KLA9JNw0fr

[Ahmad Raza Khan @kbash09]

Exactly the bet we are making at Adaptive ML by building a platform that lets you post-train LLMs quickly

https://t.co/jnsz1z5PbW

[Alex Stauffer @alexstauffer_]

We post-trained a 3B model with RL to beat Opus on spreadsheet retrieval. Faster, cheaper, more accurate.

- If a piece of your agent loop is narrow, verifiable, and highly repeatable, a tiny trained model might beat the frontier.
- The application layer is still early and new verticals are opening fast. Cheap domain specialists orchestrated by a frontier model that only spends tokens on judgment is a bet worth watching.

[Ahmad Raza Khan @kbash09]

@waqasmakhdum @demian_ai Yep, longer term cost and latency take center stage once the desired accuracy is reached

[Ahmad Raza Khan @kbash09]

@jun_song Yep, full model card or it didn’t happen