Gateway v3.0: Multi-Region Routing Lands
Gateway v3.0 ships with multi-region routing, request coalescing, and a 40% latency reduction across all supported providers.
Read more
Meridian Research
Blog
Deep dives into model performance, infrastructure engineering, and the decisions behind Meridian's product direction.
Codex Goes Multi-Region: What Changed Under the Hood
We rebuilt the Codex inference scheduler to span three AWS regions. Here's how we kept cold-start under 800ms while sharding model weights.
Read more
Our Pricing Philosophy: Why We Don't Do Credits
Tokens are a commodity. Credits are a dark pattern. We charge flat per-1M-token rates with no expiry, no tiers, and no surprises.
Read more
GPT-5.4 Review: Strengths, Weaknesses, and Where It Fits
OpenAI's mid-cycle refresh brings a 32K context window and faster tool-use. We ran it against our internal eval suite — here's the scorecard.
Read more
Evaluating Image Generation Models: A Framework
We tested DALL·E 4, Imagen 3, and Stable Diffusion 4 across prompt adherence, typography, and multi-character composition. One clear winner emerged.
Read more
Claude vs Gemini for Code: A 12-Task Shootout
We put Claude 4 and Gemini 2.5 Pro through 12 real-world coding tasks — from refactoring a Rust parser to debugging a distributed lock. The gap surprised us.
Read more