Find published research by company, product, platform or technology.
Showing 1–20 of 181 dossiers
The benchmark runs agents against Prometheus, Loki, Tempo and Grafana, checks actual outcomes across three attempts, and makes the task and grading code inspectable. Synthetic tests and vendor-run leaderboard scores still need local validation.
Grafana Pages goes beyond dashboard panels: prompt-built interactive pages can use connected telemetry and invoke external tools via MCP. It is a public Grafana Cloud preview, not a self-managed release.
At ObservabilityCON 2026, Grafana opened paid plugin distribution to verified partners. Cloud customers can trial and subscribe to integrations on their Grafana bill; partner maintenance and security requirements are part of the model.
The existing coverage-first mode keeps more query detail at the cost of possible series growth. Cost-first applies all reductions but only restores detail within the same series budget; it is a guardrail against metric-cardinality spikes, not a guarantee of a fixed bill.
The observe–test–release loop now has explicit economics: Free and Pro include 30,000 captured generations and 25 million system-initiated AI tokens per month; Pro overages start at $1.50 per 1,000 generations and $2 per million LLM Eval/Guard tokens, while ordinary telemetry is billed separately.
Kimi K3.1 Pro ships via API at $2 input/$10 output per million tokens, with a 2–3x faster Highspeed tier. Independent Artificial Analysis testing scores 55 on its Intelligence Index and 67% on Terminal-Bench 4.0. Open weights remain promised, not released.
Vendor-run live-store testing exposes the trade-off between containment, quality and response speed. Store configuration matters, and the evaluator's commercial conflict requires care.
Android Bench 2.0 moves coding-agent evaluation away from small repository fixes toward dependency upgrades, app builds, migrations and other jobs that can take a human engineer days. The results expose a much larger reliability gap than short-task benchmarks—and show that the agent harness can materially change cost and outcome.
The useful finding is not that BRIN is bad: it is that a table can still report strong column correlation while update churn has destroyed the physical range locality BRIN actually depends on. DeepSQL published the full Docker/SQL harness so teams can reproduce the failure mode on their own workloads.
Stalwart 1.0.1 fixes local credential-file permissions and a tenant isolation bug, but changes Prometheus/OTLP metrics and leaves some migration WebUI cache pitfalls.
The new Turso 0.8 results add a concrete performance datapoint to the earlier MVCC preview, but benchmark workloads, transaction conflicts and production durability still require independent verification.
Jalapeño is working first-party silicon rather than a roadmap item, and OpenAI now says AI itself materially accelerated the design process. The distinction still matters: tape-out means the design was finalized for manufacturing; it does not mean fleet-scale production qualification or API deployment is complete.
The latest private-SaaS deal-size benchmark shows median ACV moving down, with bootstrapped companies at $18,643 versus $39,880 for equity-backed peers. For small SaaS operators, the useful question is whether larger contracts improve retention and economics enough to justify the longer sales motion.
Private SaaS teams now have a fresher efficiency baseline: median ARR per employee rose to $141,125, and bootstrapped businesses lead equity-backed peers on the metric across company sizes. The same survey family shows bootstrapped $3M–$20M SaaS companies growing more slowly but generally operating with stronger cost discipline.
The strongest signal in Produktly’s 2026 onboarding dataset is not a universal target but a set of usable baselines: median tour completion was 29%, 1–2-step tours completed far more often than 9+ step tours, in-app NPS response rates were low, and announcement attention was heavily front-loaded. The report explicitly discloses sample and causal limitations.
The useful part of Smaug Agentic is not another frontier-style benchmark claim. Abacus.AI is publishing a drop-in Kimi K3 derivative that targets a specific production failure mode in coding agents: long runs that burn the reasoning budget without converging. The weights and model card are public, but the training data is not disclosed and the benchmark gains remain vendor-produced.
The new event storage format is faster in PostHog's benchmarks, but raw $set instructions will expire after 60 days and feature-flag calls after 90. Most ordinary event and person data is unaffected.
Bun 1.4.4 introduces bun fmt, accelerates bun check and preserves Yarn Berry lockfiles during migration. Teams should test formatter output and unsupported plugins before replacing Prettier in CI.
The PyTorch 2.15 RC changes what Apple-silicon developers can run locally: FP8, faster MPS decoding, new GPU operators and a video VAE memory fix. The reported 5–6.5x gain is not universal.
Cloudflare and VoidZero are developing tsrs under Oxc with the TypeScript team. The early Rust checker claims up to 8x speed and 70% lower memory than TypeScript 7.0 under matched worker counts, but is not a drop-in stable compiler yet.