Best AI Model for Coding (June 2026): 12 Models Ranked by SWE-bench Pro Score and Cost per Task

Best AI Model for Coding: Quick Answer (June 2026)

Every ranking page shows you a benchmark column or a price column. Almost none divide one by the other. This page does both: SWE-bench Pro scores from Scale's standardized leaderboard, official per-token prices, and the output-dollar cost per solved benchmark point. Updated June 9, 2026, the day Claude Fable 5 went GA.

12 Models Ranked: SWE-bench Pro x Price x Cost per Solved Point

The table below uses Scale's SEAL public-set leaderboard, which runs every model through the same standardized scaffolding on SWE-bench Pro (1,865 tasks, 41 professional repositories). The last column divides official output price by score: dollars of output tokens per benchmark point solved. Lower is more cost-effective.

Three things fall out of the combined view. GPT-5.4 wins on score and is competitive on cost per point. Haiku 4.5 solves 67% as many tasks as gpt-5.4 at a third of its output price, making it the cost-per-point leader at roughly $0.13 per point. And Opus 4.6, the top Claude entry Scale has tested, pays a 2x cost-per-point premium over gpt-5.4 for 7.2 fewer points on this harness, which is exactly why Anthropic publishes its own numbers (covered below).

SWE-bench Verified Leaderboard (June 2026)

SWE-bench Verified is older, Python-only, and partially contaminated, but it is still the number every launch post quotes. The June 2026 top ten, per the llm-stats tracker:

Two structural facts in this chart. First, the Claude line pulled away from the pack: Fable 5's 95.0% is 6.4 points above Opus 4.8 and 14.4 above the 80-percent cluster. Second, the 80-percent cluster is now half open weights. DeepSeek-V4-Pro-Max (80.6%) ties Gemini 3.1 Pro exactly, and you can download its MIT-licensed weights.

Which Claude Model Is Best for Coding?

Anthropic ships five current coding-relevant models. Exact API IDs, prices, and the decision logic:

One model to ignore: Claude Mythos 5 (93.9% SWE-bench Verified as Mythos Preview) is the same underlying model as Fable 5 with safety classifiers lifted, available only to approved Project Glasswing partners. It is also suspended as of June 12, 2026. There is no self-serve access and no active availability, so it is not a practical coding pick.

Claude Opus 4.8 vs GPT-5.5: The $5-Input Flagships

Both current flagships cost $5/M input. The output gap ($25 vs $30) and the benchmark splits decide it:

Opus 4.8 wins repo-scale software engineering (69.2% vs 58.6% SWE-bench Pro on vendor harnesses) and knowledge work (1890 vs 1769 GDPval-AA Elo). GPT-5.5 wins terminal-native workflows (78.2% vs 74.6% Terminal-Bench 2.1) and ships through Codex, where a $20/mo Plus plan covers 15-80 messages per 5-hour window without per-token billing. If your work lives in a CLI agent, Codex pricing changes the math; if it lives in long-horizon repo edits, Opus 4.8 does.

Open-Source Models: 80% SWE-bench Verified at a Tenth of the Price

The open-weights frontier moved twice this spring. DeepSeek V4 (April 24, 2026) put an MIT-licensed model at 80.6% SWE-bench Verified, and MiniMax M3 (May 31, 2026) matched it at 80.5% with a 1M context window. Official API prices and self-host terms:

The arithmetic that matters: MiniMax M3 produces output at $1.20/M against Opus 4.8's $25/M, a 20.8x gap, while trailing it by 8.1 points on SWE-bench Verified. DeepSeek V4 Flash sets the absolute floor at $0.28/M output with a 1M-token context and 384K max output. The price gap is mostly model size, not provider margin, which is the core fact of how AI inference is priced. For teams with data-sovereignty requirements, DeepSeek V4's MIT license means the 80.6% model is self-hostable outright. Deeper coverage on the best open-source coding model page.

Best AI Model for Coding at $0

Four paths to real coding capability without a credit card, as of June 2026:

The honest framing: free tiers are for evaluation and light use. DeepSeek V4 Flash's paid API at $0.14/M input ($0.0028/M on cache hits) and $0.28/M output is close enough to zero that most teams skip self-hosting unless data cannot leave their network.

Why Vendor Scores Run 20 Points Above Scale's Leaderboard

Anthropic reports Fable 5 at 80.3% on SWE-bench Pro. Scale's standardized leaderboard tops out at 59.10% (gpt-5.4). Both numbers are real. The difference is the harness: Scale runs every model through identical scaffolding; vendors run their own tuned agent stacks.

The practical conclusion has not changed since 2025: the scaffold around the model accounts for more variance than swapping frontier models. Before paying a 2x token premium, fix retrieval, context management, and tool design. Subagent architecture and context engineering move scores more than model choice does.

Per-Task Routing: Which Model for Which Job

The most cost-effective setups in June 2026 route by task, not by loyalty. Numbers-backed defaults:

Cost levers that apply across routes: Anthropic's Batch API is 50% off input and output, prompt-cache reads are 0.1x base input, and DeepSeek cache hits drop input to $0.0028/M. A routing setup that pins 80% of traffic to Haiku 4.5 or DeepSeek V4 Flash and reserves Opus 4.8 / Fable 5 for the hard 20% typically beats any single-model subscription. Doing the split automatically needs a classifier: Morph's Router scores each prompt by difficulty and domain in ~180ms and returns the cheapest capable model, and Claude Code Router makes that per-request routing concrete inside the terminal agent. Which model your editor agent uses matters too; see Claude Code models for harness-side defaults.

Frequently Asked Questions