Introducing the Open-Weight League

Today we're launching The Open-Weight League — a live league table for open-weight AI models, published at valarian.com/owl and updated every Friday.

Twenty clubs. Four stats — intelligence, coding, speed, openness. One blended score that sets the table, with a frontier at the top and a relegation zone at the bottom. We built it as a form guide rather than a static list, because that's what the open-weight world actually is now: models improve, slip, and get displaced week by week, and a snapshot can't show you form.

We built it because our customers keep asking the same question: which open-weight models are actually good right now? Open weights are the sovereign choice — models you can download and run on infrastructure you control, with keys only you hold. The League is that argument, in table form. Season 2026/27 kicks off on World Cup final weekend, which felt right.

View the League table →

The short version

Every Friday, we rank the 20 best open-weight models in the world. The scores come from Artificial Analysis, an independent AI benchmarking organisation. The ranking formula is ours, and it's published below — in full. Whether a model runs in Valarie has no bearing on its position. That's the whole game.

Who gets in

The League is open weights only. To qualify, a model must have downloadable weights you can run on infrastructure you control — your cluster, your cloud, your air-gapped estate.

Concretely, a club qualifies when:

• Its weights are publicly downloadable (Hugging Face or equivalent),
• Its licence permits self-hosted deployment,
• It is benchmarked by Artificial Analysis, so scores exist for every column.

Closed, API-only models are not eligible, however capable. This isn't a ranking of all AI; it's a ranking of the AI you can actually own. Closed models play a different sport.

From all qualifying models, the top 20 by League Points make the table each Matchday. Where a lab has several qualifying models, each model is its own club. Labs can, and do, occupy multiple places in the table.

Where the numbers come from

Every performance score in the League comes from Artificial Analysis, an independent AI benchmarking organisation. They evaluate models continuously across intelligence, coding, and speed, and publish their methodology in full.

We chose an external source deliberately. We don't run our own evaluations, we don't adjust their numbers, and we can't nudge a score. You don't have to trust us on the data, only on the formula. And the formula is right here.

Reading the table — the four stats

Every column answers a different question a deployer actually asks: how smart is it (INT), can it do real engineering (COD), can I run it at production speed (SPD), and how much of it do I truly get (OPEN). All four come directly from Artificial Analysis, which means every number in the table is third-party. The only thing Valarian adds is the blend.

INT — Intelligence Index

Artificial Analysis' composite measure of general capability, the model's overall rating. It answers one question: how smart is this model across the full range of real work?

What goes into it (Intelligence Index v4.1 — nine evaluations, weighted by category):

• Agentic tasks (~34%) — the largest share: can the model operate in a loop, use tools, and complete long-horizon work? Includes GDPval-AA (real tasks from 44 occupations, where the model must produce actual deliverables: spreadsheets, memos, presentations), τ³-Banking (multi-step customer-service workflows), and Terminal-Bench (working in a real command-line environment).
• Coding (~24%) — evaluations like SciCode, testing whether the model can write working scientific code.
• Scientific reasoning (~24%) — the hardest academic benchmarks: GPQA Diamond (PhD-level science questions), Humanity's Last Exam, CritPt (frontier physics).
• Knowledge & reliability (~18%) — including AA-Omniscience, which doesn't just reward correct answers: it penalises hallucinations, so a model that confidently makes things up scores worse than one that knows what it doesn't know. AA-LCR tests long-context reasoning: holding and using large amounts of information at once.

In short: INT rewards models that can reason, act, code, and know their own limits, rather than models tuned to ace one famous benchmark.

COD — Coding Index

Artificial Analysis' composite score for software engineering, broken out as its own column because coding is the capability production deployments ask for most. It answers: can this model do real engineering work?

What it measures: performance across software engineering and code generation benchmarks — writing new code, editing and debugging existing code, and completing engineering tasks the way a working developer would. It's related to INT (coding evaluations contribute there too) but isolated here so you can spot the specialists: a model can be a mid-table all-rounder and still be a top-5 engineer, or vice versa. If you're choosing a model to put in a development workflow, this is your column.

SPD — Output Speed

How fast the model actually generates, measured as median output tokens per second across serving providers, as benchmarked by Artificial Analysis.

Why median, and why across providers: the same open-weight model can run at very different speeds depending on who serves it and on what hardware. The median across providers reflects what you can realistically expect in practice, rather than one provider's best-case demo. And why it's in the table at all: throughput determines whether a model is usable in production. In agentic workflows, where a model might generate thousands of tokens per task, a slow model is a bottleneck whatever it costs. Capability you can't run at production speed is capability on paper.

OPEN — Openness Index

Artificial Analysis' Openness Index, scored out of 18. It measures how much of itself a model actually publishes — not how good the model is, but how yours it can be.

Four components, each scored 0–3 against defined openness "archetypes," summed to a maximum of 18:

• Model availability — are the weights downloadable, and under what licence? A permissive licence (use it, modify it, deploy it commercially) scores higher than a restrictive one.
• Pre-training data transparency — does the lab disclose what the model was trained on?
• Post-training data transparency — the same for fine-tuning and preference data. (The two data components are averaged together in the total.)
• Methodology transparency — is the build documented? Technical reports, training recipes, tooling.

Why this deserves a column: two models can both be "open weights" and sit at opposite ends of this scale. One ships weights, data recipes, and a full technical report under a permissive licence; another drops weights under a restrictive licence and explains nothing. For anyone deploying models on their own infrastructure, that gap is the gap between something you can audit, reproduce, and truly own, and a black box you happen to host. No model evaluated at the Index's launch reached the full 18: full openness is genuinely rare, which is exactly why we award points for it.

How League Points are calculated

PTS is the blended score that sets the table — the one editorial layer we add on top of Artificial Analysis data:

PTS = INT × 40% + COD × 25% + SPD × 15% + OPEN × 20%

The four stats live on different scales — INT and COD are index scores, SPD is raw tokens per second, OPEN is out of 18 — so each component is normalised to a common 0–100 scale before weighting.

Why these weights

We could have hidden the weighting. Instead, here's the argument:

• Intelligence, 40% — it's what you're buying. General capability is the product; everything else is a modifier.
• Coding, 25% — the capability production deployments ask for most, and the best single proxy for a model doing real work.
• Openness, 20% — weighted above speed, on purpose. This is an open-weight league: a model that publishes more of itself — weights, licence, data, methods — is structurally more sovereign, more auditable, and more yours. That's worth points.
• Speed, 15% — the smallest weight, but not zero: a model you can't run at production throughput isn't a real option, whatever its scores.

Reasonable people can defend different weights. These are ours, they're published, and they apply identically to every club.

Stability rules

• The formula never changes mid-season. Weights can be revised between seasons only, announced in advance, and archived so past tables remain comparable.
• When Artificial Analysis revises an index version (e.g. Intelligence Index v4.1 → v5), we adopt the new version at the next Matchday, say so in the changelog, and never mix versions within a single table.
• If a score is corrected upstream by Artificial Analysis, we recompute at the next Matchday. We don't retroactively edit published tables; the archive shows what was known at the time.

The zones

FRONTIER (1st). Top of the table. The best open-weight model in the world right now. QUALIFICATION (2nd–5th). The chasing pack: models with a credible claim to the top spot. RELEGATION (18th–20th). On the bubble. A qualifying model outside the top 20 can displace them at any Matchday.

Movement arrows (▲▼) show position change versus the previous Matchday. A dash means the club held its position.

Cadence

The table updates every Friday — one Matchday per week, computed from Artificial Analysis' latest published data at the time of ranking. Season 2026/27 runs July 2026 to June 2027.

The Matchday awards

Three awards accompany every Matchday. The first two are pure maths; the third belongs to the terraces.

MVM — Most Valuable Model. The biggest League Points gain versus the previous Matchday. Tiebreaker: biggest climb in position. If no model gains points, no MVM is awarded. The honour has to be earned.

Rookie of the Matchday. The highest new entry into the top 20, awarded when a model breaks in for the first time. No new entry, no Rookie: the top 20 held.

The Supporters' Award. The fans' choice. Every Wednesday we open a vote on X with four nominees, selected by a fixed rule: the current Frontier holder, the two biggest climbers of the previous Matchday, and the latest Rookie (or an editorial wildcard when there is none). Voting closes Friday morning; the winner and vote share are announced with the table. Votes are counted from the official poll only, and the result is final.

Awards are honours, not points: winning any award — including the Supporters' Award — has zero effect on League position. The table answers to the data; the awards answer to the week's story and to the crowd.

Independence

Some clubs in the League are validated in Valarie, our sovereign agent, and available to run in ACRA, our platform. This has zero influence on the rankings: positions are computed from Artificial Analysis' published data and the weighting formula above, and nothing else. No pay-to-play, no partner boosts, no exceptions. We built the League because we believe open weights are the sovereign choice. The table is the argument, and it only works if it's honest.

Questions we get

Why isn't [closed model] in the table? Because you can't download it. The League ranks models you can own and run yourself.

A model's benchmark score looks different elsewhere. Different evaluators, different harnesses, different numbers. We use Artificial Analysis exclusively, for consistency; their methodology is linked above.

Can a lab submit a model? No submission needed: if Artificial Analysis benchmarks it and the weights are downloadable, it's automatically eligible.

Who decides all this? The formula and rules on this page decide. We wrote them, we publish them, and we're bound by them like everyone else.

The table is live now at valarian.com/owl. See you Friday.

Source: Artificial Analysis (artificialanalysis.ai). The Open-Weight League is an editorial product of Valarian. Questions or corrections: get in touch.