MODEL FRONTIER
AI models
Track ChatGPT, Claude and the wider foundation-model market across releases, benchmarks, APIs, open weights and deployment choices.
EDITOR'S SELECTION
Latest analysis
OpenAI launches GPT-6 Astra for ChatGPT Work, Codex and the API
OpenAI's newest flagship model targets computer use and professional workflows, with API pricing beginning at $10 per million input tokens and $50 per million output tokens.
· Ravi Kapoor · 10 min readLiquid AI releases draft models for faster local inference
Liquid AI's LFM2.5-DSpark checkpoints use speculative decoding to accelerate several LFM2.5 models. The vendor reports large gains, but teams should reproduce them on their own hardware and workloads.
· Elena Morris · 8 min readWhat speculative decoding changes for local AI agents
Draft-and-verify inference can make an agent feel faster without replacing its target model. The benefit is workload-specific, and poor evaluation can hide memory and reliability costs.
· Ethan Brooks · 9 min readWhat is ChatGPT in 2026? A practical guide to features, workflows and limits
ChatGPT has grown from a text chatbot into a workspace for search, files, images, voice and longer projects. This guide explains what the product does, where it helps and where human verification still matters.
· Priya Raman · 11 min readWhat is artificial intelligence? A systems-level explanation without the hype
Artificial intelligence is not one technology or one level of capability. It is a family of systems that infer outputs from inputs—and whose value depends on data, objectives, interfaces and human oversight.
· Priya Raman · 11 min readHow does AI work? From training data to predictions, generation and feedback
Modern AI systems learn statistical relationships from examples, turn new inputs into outputs and improve through evaluation and feedback. The important details lie in the objective, the data and the surrounding controls.
· Priya Raman · 12 min readGPT-5.6 arrives across ChatGPT, Codex and the OpenAI API
OpenAI's July flagship release brings one model family to consumer chat, coding agents and the API, followed by substantial price cuts for its Luna and Terra variants.
· Ravi Kapoor · 7 min readClaude Sonnet 5 brings a more agentic default model to Claude Code
Anthropic's new Sonnet model targets coding, tool use and autonomous work while keeping a lower price point than its Opus-class frontier models.
· Ethan Brooks · 8 min readFLUX 3 expands Black Forest Labs from images into video and audio
The early-access model jointly learns from images, video and sound, signaling a move from specialist image generation toward a multimodal foundation for visual intelligence.
· Noah Chen · 7 min readThe model race is becoming local, multilingual and sovereign
Sarvam’s open 30B and 105B models ignited a broader conversation about who builds a country’s AI layer—and whether global benchmarks capture the languages and workflows that matter locally.
· Ravi Kapoor · 7 min readWhat an open-model breakthrough actually changes
A strong benchmark score can earn attention overnight. The harder questions are whether a model can be inspected, deployed affordably and trusted on the tasks that matter outside a leaderboard.
· Ravi Kapoor · 8 min readReasoning models need different prompts—and different tests
Longer internal computation can improve difficult analysis and coding, but it also changes latency, cost and the way teams should write instructions and evaluate results.
· Priya Raman · 8 min readSmall models are moving AI from the cloud onto devices
On-device language models trade frontier-scale breadth for low latency, privacy and zero per-request cloud cost. That makes them a different product primitive, not merely a smaller benchmark entry.
· Ravi Kapoor · 9 min readHow to evaluate a multimodal model
Models that read text, images, audio and video promise one interface for many kinds of work. A useful comparison tests perception, reasoning, latency and failure behavior separately.
· Priya Raman · 9 min read