The Signal
The AI stories that matter — and the tools worth your time
This Week in AI — What Actually Matters
Claude 4 Lands — and Agentic AI Finally Starts to Deliver
Anthropic shipped Claude 4 Sonnet and Opus with a dramatically improved agentic architecture. The real story isn’t benchmarks — it’s that multi-step tool use is now reliable enough that developers are deploying it in production. Claude Sonnet at $3/M input tokens is the most compelling value play in frontier AI right now. The 200K context window, the honest documentation, and writing quality noticeably ahead of GPT-4o on prose-heavy tasks — this is the one to watch.
GitHub Copilot Goes Multi-Model — OpenAI Codex Returns as an Agent
Microsoft opened Copilot to Claude, Gemini, and o3 alongside GPT-4o. OpenAI’s new Codex runs tasks asynchronously — submit a task, come back to a pull request. Compelling for teams with repetitive engineering work.
ByteDance’s Doubao: The Quiet Giant Most Europeans Are Ignoring
Doubao Pro and Lite consistently outperform their weight class on coding and reasoning. At $0.14/M tokens for Lite, cost-per-quality is unmatched — if your data policy permits a Chinese-hosted endpoint.
Gemini 2.5 Pro — the Dark Horse in the Context Window Wars
1M tokens natively, and it actually performs across the full window. Researchers report results matching chunking-based RAG pipelines — without the complexity. API still in preview but worth getting on your radar.
Free Tools Worth Having Right Now
Run Llama 3, Mistral, Qwen, Gemma, and dozens more entirely on your own machine. Zero API cost, zero data leaving your hardware. If you have a modern GPU (8GB+ VRAM) or Apple Silicon, this is the most important tool you’re not yet using.
200+ LLMs, one API key. A rotating list of completely free models includes Llama 3.1 405B, Gemma 3 27B, and several Mistral variants. One account, one key, genuinely capable models at no cost. Perfect for finding your model before going direct.
Full access to Gemini 2.5 Pro with its 1M context window — free in the playground. Rate-limited but entirely sufficient for experimentation and document analysis. No credit card required.
Daily access to Claude 3.5 Sonnet with generous limits. Quality-per-message among the highest of any free AI product. Worth having as a second model even if you’re already paying for something else.
The Honest LLM API Guide — May 2026
| Provider / Model | Best For | In $/M | Out $/M | Verdict |
|---|---|---|---|---|
| Claude 3.5 SonnetAnthropic | Agentic, writing, long context | $3.00 | $15.00 | Top Pick |
| Claude 3 HaikuAnthropic | Volume, classification, speed | $0.25 | $1.25 | Best Value |
| GPT-4o + CodexOpenAI / GitHub | Coding, IDE, PR automation | $5.00 | $15.00 | Dev First |
| Gemini 2.5 ProGoogle AI / Vertex | Long docs, multimodal, research | $1.25 | $10.00 | Context King |
| Doubao Pro / LiteByteDance (Volces) | Bulk, cost-critical workloads | $0.14 | $0.28 | Cheapest |
| Ollama (local)Your hardware | Privacy, offline, zero cost | £0 | £0 | Free Forever |
| OpenRouteropenrouter.ai | Model switching, free access | Varies | Varies | Most Flexible |
Our preferred API for anything requiring actual thinking — long-form writing, complex instruction following, multi-step agentic work. Writing quality noticeably ahead of GPT-4o for prose-heavy tasks. The documentation is the most honest in the industry. Watch for: output pricing at $15/M adds up fast — use Haiku ($1.25/M) for anything that doesn’t need Sonnet quality.
GitHub Copilot is the most deeply integrated coding assistant at £10/month — lives in VS Code, JetBrains, Neovim. The new Codex cloud agent runs tasks asynchronously: submit a task, come back to a pull request. Compelling for teams with repetitive engineering work. Watch for: $15/M output and per-task Codex billing at scale.
The most capable model for ingesting large documents or codebases in one shot. 1M tokens that actually performs across the full window — more than can be said for earlier long-context attempts. Google AI Studio free tier lets you test properly before paying. Watch for: enterprise-heavy documentation creates higher onboarding friction than Anthropic or OpenAI.
Doubao Lite 32K at $0.14/M input is among the lowest-cost options that produces coherent output. OpenAI-compatible API — migration is straightforward. Watch for: data residency is in China. Hard blocker for EU user data or sensitive commercial IP. English documentation is patchy; customer support non-existent in Western time zones.
One command to download, one to run. OpenAI-compatible REST API locally — your existing tooling connects with a single URL change. Library includes Llama 3.1, Mistral Nemo, Qwen 2.5, DeepSeek, Gemma 3, Phi-4. Watch for: quality is hardware-dependent. 8B on GPU = good. 7B on CPU = frustrating. 70B needs 40GB VRAM — server territory.
One API key, one billing account, access to models from Anthropic, OpenAI, Google, Meta, Mistral, and dozens more — including smaller labs you’d never find otherwise. Provider pricing plus a small markup. Free-tier models included. Watch for: a middleman adds latency and a point of failure. Go direct at production scale with a single model.
Also From AI Maestro — Related Reading
Guide · AI Coding Tools
Claude Code vs GitHub Codex vs Cursor: The Honest 2026 Comparison
Three AI coding assistants, one honest verdict. We tested each on real tasks — not synthetic benchmarks.
Read the guide →
Guide · Infrastructure
GPU Rental vs LLM API vs Cloud Hosting: Which Actually Makes Sense?
The infrastructure decision most developers get wrong. Real numbers, real trade-offs — no affiliate padding.
Read the guide →
Review · Tools
Ollama Cloud Review 2026: Is It Actually Worth It?
We ran Ollama Cloud through its paces. Here is what the marketing doesn’t tell you — and when local is still the better call.
Read the review →