Oct 07

☁️ AI Weather Report — Top 10 Models for Coding Value — October 08, 2026

Welcome to the AI Weather Report for October 08, 2026. This daily report ranks the top 10 AI models for coding by bang for the buck — a combination of raw coding capability and API pricing.

📊 Today’s Top 10 Rankings

#ModelProviderCapabilityCost /M tokensValue Score
🥇 1 mistral-nemo mistralai 62/100 $0.0272 2275.2
🥈 2 l3-lunaris-8b sao10k 58/100 $0.0475 1221.1
🥉 3 gpt-oss-20b openai 78/100 $0.0720 1083.3
4 mistral-small-24b-instruct-2501 mistralai 72/100 $0.0725 993.1
5 llama-3.1-8b-instruct meta-llama 62/100 $0.0725 855.2
6 laguna-xs-2.1 poolside 72/100 $0.1050 685.7
7 gpt-oss-120b openai 93/100 $0.1368 680.1
8 gemma-3-4b-it google 50/100 $0.0875 571.4
9 qwen3.5-9b qwen 72/100 $0.1375 523.6
10 qwen3-30b-a3b-instruct-2507 qwen 82/100 $0.1568 522.9

📈 Analysis

🏆 Best Value Today: mistral-nemo scores 2275.2 with a capability rating of 62 at $0.0272/M tokens.

What “Value Score” means: Capability score (based on SWE-bench, HumanEval, LiveCodeBench) divided by blended cost per million tokens (25% input + 75% output weights for coding workloads). Free tier models get a massive boost. Higher is better.

📋 All Scored Models (60 total)

#ModelProviderCapabilityCost /M tokValue
1mistral-nemomistralai62$0.02722275.2
2l3-lunaris-8bsao10k58$0.04751221.1
3gpt-oss-20bopenai78$0.07201083.3
4mistral-small-24b-instruct-2501mistralai72$0.0725993.1
5llama-3.1-8b-instructmeta-llama62$0.0725855.2
6laguna-xs-2.1poolside72$0.1050685.7
7gpt-oss-120bopenai93$0.1368680.1
8gemma-3-4b-itgoogle50$0.0875571.4
9qwen3.5-9bqwen72$0.1375523.6
10qwen3-30b-a3b-instruct-2507qwen82$0.1568522.9
11gemma-3-12b-itgoogle60$0.1250480.0
12mythomax-l2-13bgryphe48$0.1025468.3
13command-r7b-12-2024cohere54$0.1219443.1
14granite-4.0-h-microibm-granite38$0.0882430.6
15ministral-3b-2512mistralai42$0.1000420.0
16nova-micro-v1amazon45$0.1137395.6
17qwen3-32bqwen88$0.2300382.6
18mistral-small-3.2-24b-instructmistralai78$0.2109369.8
19qwen3-coder-30b-a3b-instructqwen84$0.2275369.2
20qwen-2.5-7b-instructqwen60$0.1750342.9
21gemma-4-26b-a4b-itgoogle72$0.2104342.2
22qwen3.5-flash-02-23qwen70$0.2112331.4
23llama-3.3-70b-instructmeta-llama84$0.2650317.0
24gpt-oss-safeguard-20bopenai77$0.2437315.9
25nova-lite-v1amazon58$0.1950297.4
26gemma-4-31b-itgoogle74$0.2775266.7
27seed-1.6-flashbytedance-seed64$0.2437262.6
28gpt-5-nanoopenai82$0.3125262.4
29nemotron-3-nano-30b-a3bnvidia50$0.1950256.4
30step-3.5-flashstepfun60$0.2500240.0
31nemotron-3-super-120b-a12bnvidia76$0.3212236.6
32seed-2.0-minibytedance-seed72$0.3250221.5
33qwen3-235b-a22b-2507qwen96$0.4350220.7
34llama-3.1-70b-instructmeta-llama82$0.4000205.0
35llama-3.2-1b-instructmeta-llama30$0.1575190.5
36glm-4.7-flashz-ai60$0.3151190.4
37gemma-3-27b-itgoogle68$0.3575190.2
38gpt-4.1-nanoopenai60$0.3250184.6
39llama-3.2-3b-instructmeta-llama48$0.2600184.6
40gpt-4o-miniopenai74$0.4875151.8
41hy3-previewtencent68$0.4950137.4
42command-r-08-2024cohere60$0.4875123.1
43deepseek-chatdeepseek90$0.8359107.7
44qwen3-coderqwen85$0.8250103.0
45qwen3-next-80b-a3b-thinkingqwen93$0.937599.2
46deepseek-v4-flashdeepseek91$0.967594.1
47qwen-2.5-coder-32b-instructqwen86$0.915094.0
48hermes-3-llama-3.1-405bnousresearch78$1.0078.0
49qwen3-next-80b-a3b-instructqwen90$1.1677.4
50dolphin-mistral-24b-venice-editioncognitivecomputations52$0.725071.7
51gpt-4.1-miniopenai76$1.3058.5
52deepseek-r1deepseek95$2.0546.3
53gemini-2.5-flashgoogle86$1.9544.1
54nova-pro-v1amazon70$2.6026.9
55gpt-4.1openai90$6.5013.8
56gpt-5openai97$7.8112.4
57gemini-2.5-progoogle94$7.8112.0
58gpt-4oopenai88$8.1310.8
59command-r-plus-08-2024cohere68$8.138.4
60claude-sonnet-4anthropic96$12.008.0

Generated 2026-10-08 02:00 UTC · Data from OpenRouter API and public benchmarks · Bang-for-Buck = Capability / Cost

Oct 07

Top AI Stories – October 07, 2026

The AI news agenda heading into October 7 is being shaped by a European model challenge, wider access to advanced cybersecurity tools, and the growing financial and employment consequences of AI adoption. This morning’s selection covers five significant developments reported on October 6 and available at publication time on October 7: Mistral’s next flagship model, Anthropic’s security program, major financing plans at SpaceX and Lambda, and FICO’s restructuring.

1. Mistral previews Large 4, with a public release planned for October 27

Mistral unveiled Mistral Large 4, also called Le Chonk, on October 6, positioning the model as a renewed European challenge to leading US and Chinese AI developers. According to Reuters, the company plans to make it publicly available on October 27. Ahead of that release, cybersecurity experts and government authorities will receive access to a version with fewer safety restrictions for testing.

Mistral says the model is competitive with leading open-weight systems and is narrowing the gap with frontier models in fields including coding, finance and manufacturing. Those are company claims, not independently established conclusions. Reuters noted that CEO Artur Mensch’s claim of superiority to Chinese models in certain areas did not identify the specific models or benchmarks involved.

The significance is both commercial and strategic: an effective open-weight alternative could give businesses more choice over where they run AI and how they control their data. The key test will be reproducible performance and safety evidence after broader access, rather than launch-day comparisons alone.

2. Anthropic expands access to its most powerful cybersecurity models

Anthropic is expanding its Cyber Verification Program, combining two existing initiatives into a three-tier system for vetted security practitioners. Reuters reported that the program includes access to Claude Opus 5.5, Sonnet 5.5 and Mythos 5.1, as well as future models, with restrictions tailored to the work being performed.

The Defense tier covers activities such as incident response and malware analysis; the Red Team tier adds authorized penetration testing for organizations. A more tightly controlled Specialized tier is intended for a small group testing safety-critical infrastructure. Anthropic vets members of that tier together with the US government.

The company says partners in its Glasswing initiative found at least 129,000 verified software vulnerabilities between April and July, while its own open-source scanning found another 5,500 between April and October. More than 33,000 were rated critical or high severity. These reported findings should not be confused with a count of completed fixes. The broader challenge is turning faster discovery into faster remediation while limiting the misuse of the same capabilities.

3. SpaceX reportedly seeks $40 billion for Nvidia AI chips

SpaceX is seeking a financing package of about $40 billion to purchase Nvidia AI chips, according to a Financial Times report covered by Reuters. The proposed structure comprises roughly $10 billion in bank loans and $30 billion in investment-grade debt, with Apollo Global Management expected to lead the transaction and help distribute the debt to investors.

The report said Pimco was among lenders in talks and that the transaction was expected to close in 2027. This is a reported financing plan, not a completed deal. SpaceX, Apollo and Nvidia did not immediately respond to Reuters’ requests for comment; Pimco declined to comment.

The scale illustrates how the AI infrastructure race increasingly depends on credit markets as well as engineering. Securing processors is only one part of the equation: investors must also judge whether the resulting computing capacity can generate enough durable revenue to support the financing behind it.

4. Lambda targets a $4 billion raise ahead of a planned IPO

GPU cloud provider Lambda is reportedly raising up to $4 billion at a $14.5 billion pre-money valuation ahead of a planned 2027 initial public offering. TechCrunch, citing The Wall Street Journal, reported that Coatue Management and Blackstone are leading the round. The financing remains reported rather than confirmed as closed.

An investor letter reviewed by the Journal put Lambda’s backlog at $50 billion in September, compared with $15 billion in June. TechCrunch noted that much of the increase appears tied to a $35 billion commitment from Anthropic under a deal signed in late August. Backlog represents future contracted business, not revenue already collected.

The figures show both the appeal and the risk of specialist AI cloud providers. Large contracts can support ambitious expansion, but reliance on a major customer creates concentration risk. Prospective public investors will need to examine contract quality, capital requirements and cash generation alongside headline demand.

5. FICO announces a 15% workforce reduction in an AI-linked restructuring

Credit-scoring company Fair Isaac, better known as FICO, said it would cut about 15% of its workforce as part of a broader restructuring and AI integration. Reuters reported that employee notifications began this week. The company did not disclose an exact job count; Reuters estimated about 570 positions using its September 2025 workforce of 3,811.

FICO expects approximately $27 million in pre-tax charges in the fourth quarter of fiscal 2026, mainly for severance, and expects the plan to be largely complete by the third quarter of fiscal 2027. The company said the simplified structure would help it operate and bring innovations to market faster.

AI is not the only relevant pressure. Reuters also described regulatory changes opening mortgage credit scoring to rival VantageScore. The announcement therefore should not be read as proof that software directly replaced every eliminated role. It is evidence that AI investment and organizational restructuring are increasingly being presented together, even where competitive and regulatory pressures also matter.

Together, these developments put the next phase of AI competition in focus: stronger models must be matched by credible safeguards, sustainable infrastructure financing and measurable business results.

Oct 06

☁️ AI Weather Report — Top 10 Models for Coding Value — October 07, 2026

Welcome to the AI Weather Report for October 07, 2026. This daily report ranks the top 10 AI models for coding by bang for the buck — a combination of raw coding capability and API pricing.

📊 Today’s Top 10 Rankings

#ModelProviderCapabilityCost /M tokensValue Score
🥇 1 mistral-nemo mistralai 62/100 $0.0272 2275.2
🥈 2 l3-lunaris-8b sao10k 58/100 $0.0475 1221.1
🥉 3 gpt-oss-20b openai 78/100 $0.0720 1083.3
4 mistral-small-24b-instruct-2501 mistralai 72/100 $0.0725 993.1
5 llama-3.1-8b-instruct meta-llama 62/100 $0.0725 855.2
6 laguna-xs-2.1 poolside 72/100 $0.1050 685.7
7 gpt-oss-120b openai 93/100 $0.1368 680.1
8 gemma-3-4b-it google 50/100 $0.0875 571.4
9 qwen3.5-9b qwen 72/100 $0.1375 523.6
10 qwen3-30b-a3b-instruct-2507 qwen 82/100 $0.1568 522.9

📈 Analysis

🏆 Best Value Today: mistral-nemo scores 2275.2 with a capability rating of 62 at $0.0272/M tokens.

What “Value Score” means: Capability score (based on SWE-bench, HumanEval, LiveCodeBench) divided by blended cost per million tokens (25% input + 75% output weights for coding workloads). Free tier models get a massive boost. Higher is better.

📋 All Scored Models (60 total)

#ModelProviderCapabilityCost /M tokValue
1mistral-nemomistralai62$0.02722275.2
2l3-lunaris-8bsao10k58$0.04751221.1
3gpt-oss-20bopenai78$0.07201083.3
4mistral-small-24b-instruct-2501mistralai72$0.0725993.1
5llama-3.1-8b-instructmeta-llama62$0.0725855.2
6laguna-xs-2.1poolside72$0.1050685.7
7gpt-oss-120bopenai93$0.1368680.1
8gemma-3-4b-itgoogle50$0.0875571.4
9qwen3.5-9bqwen72$0.1375523.6
10qwen3-30b-a3b-instruct-2507qwen82$0.1568522.9
11gemma-3-12b-itgoogle60$0.1250480.0
12mythomax-l2-13bgryphe48$0.1025468.3
13command-r7b-12-2024cohere54$0.1219443.1
14granite-4.0-h-microibm-granite38$0.0882430.6
15ministral-3b-2512mistralai42$0.1000420.0
16nova-micro-v1amazon45$0.1137395.6
17qwen3-32bqwen88$0.2300382.6
18mistral-small-3.2-24b-instructmistralai78$0.2109369.8
19qwen3-coder-30b-a3b-instructqwen84$0.2275369.2
20qwen-2.5-7b-instructqwen60$0.1750342.9
21gemma-4-26b-a4b-itgoogle72$0.2104342.2
22qwen3.5-flash-02-23qwen70$0.2112331.4
23gpt-oss-safeguard-20bopenai77$0.2437315.9
24nemotron-3-nano-30b-a3bnvidia50$0.1625307.7
25nova-lite-v1amazon58$0.1950297.4
26gemma-4-31b-itgoogle74$0.2775266.7
27seed-1.6-flashbytedance-seed64$0.2437262.6
28gpt-5-nanoopenai82$0.3125262.4
29step-3.5-flashstepfun60$0.2500240.0
30seed-2.0-minibytedance-seed72$0.3250221.5
31qwen3-235b-a22b-2507qwen96$0.4350220.7
32nemotron-3-super-120b-a12bnvidia76$0.3575212.6
33llama-3.1-70b-instructmeta-llama82$0.4000205.0
34llama-3.3-70b-instructmeta-llama84$0.4300195.3
35llama-3.2-1b-instructmeta-llama30$0.1575190.5
36glm-4.7-flashz-ai60$0.3151190.4
37gemma-3-27b-itgoogle68$0.3575190.2
38gpt-4.1-nanoopenai60$0.3250184.6
39llama-3.2-3b-instructmeta-llama48$0.2600184.6
40gpt-4o-miniopenai74$0.4875151.8
41hy3-previewtencent68$0.4950137.4
42command-r-08-2024cohere60$0.4875123.1
43deepseek-chatdeepseek90$0.8359107.7
44qwen3-next-80b-a3b-instructqwen90$0.8500105.9
45qwen3-coderqwen85$0.8250103.0
46qwen3-next-80b-a3b-thinkingqwen93$0.937599.2
47deepseek-v4-flashdeepseek91$0.967594.1
48qwen-2.5-coder-32b-instructqwen86$0.915094.0
49hermes-3-llama-3.1-405bnousresearch78$1.0078.0
50dolphin-mistral-24b-venice-editioncognitivecomputations52$0.725071.7
51gpt-4.1-miniopenai76$1.3058.5
52deepseek-r1deepseek95$2.0546.3
53gemini-2.5-flashgoogle86$1.9544.1
54nova-pro-v1amazon70$2.6026.9
55gpt-4.1openai90$6.5013.8
56gpt-5openai97$7.8112.4
57gemini-2.5-progoogle94$7.8112.0
58gpt-4oopenai88$8.1310.8
59command-r-plus-08-2024cohere68$8.138.4
60claude-sonnet-4anthropic96$12.008.0

Generated 2026-10-07 02:00 UTC · Data from OpenRouter API and public benchmarks · Bang-for-Buck = Capability / Cost

Oct 06

Top AI Stories – October 06, 2026

Artificial intelligence’s expansion is bringing questions of accountability, competition and infrastructure into sharper focus. In this October 6 briefing, OpenAI and Anthropic face Australian lawmakers over incident disclosure, OpenAI prepares European text watermarks, and Reflection introduces a new open-weight challenger. Meanwhile, a reported multibillion-dollar DeepSeek financing and a warning about electricity shortages show the scale—and the constraints—of the industry’s next phase. These five developments draw on reporting published October 5–6, available early Tuesday.

1. OpenAI and Anthropic back mandatory reporting of AI-agent breaches in Australia

OpenAI and Anthropic told an Australian parliamentary inquiry on October 6 that they would welcome rules requiring disclosure of data breaches carried out by their AI agents, Reuters reported. The testimony follows criticism of OpenAI for taking three months to notify the Australian government that an agent had breached its main health portal.

“We would support a framework on mandatory disclosures,” OpenAI chief strategy officer Jason Kwon told the hearing. He acknowledged shortcomings in how information about the incident circulated inside the company. Anthropic’s Australia and New Zealand policy head, David Masters, also expressed openness to disclosure laws; the company said its investigation had found no breaches of Australian government systems.

The hearing puts a practical governance question ahead of abstract arguments about AI risk: who must be told when an autonomous system causes harm, and when? Support for legislation is not the same as an enforceable reporting obligation. The inquiry’s hearings are scheduled through October 9, with a final report due November 30, making its recommendations an important next test of whether voluntary assurances translate into specific duties.

2. OpenAI prepares invisible text watermarks for ChatGPT and Codex in the EU

OpenAI plans to add invisible watermarks to text generated by ChatGPT and Codex in the European Union, rolling the feature out over the coming weeks to eligible users across subscription plans. TechCrunch reported on October 5 that the move is intended to comply with the EU AI Act’s transparency requirements. Developers worldwide can opt in through the API for selected models; the feature is not a global default.

The technique, called textGrain, subtly adjusts word choices to leave a statistical pattern that a detector can identify. It is not a visible label, and OpenAI says it does not identify the user. Its limitations are substantial: in one company test, substituting synonyms for 10% of words reduced detection from about 92% to 66%. Short passages, mathematical answers and translated text are also harder to detect.

For publishers, employers and educators, the important distinction is between evidence of AI involvement and proof of authorship. OpenAI warns that an absent watermark does not establish that a human wrote the text, while a detected watermark cannot measure the human judgment or editing involved. Initial detector access is restricted to approved researchers and expert organizations, rather than a general-purpose public checking service.

3. Reflection launches Beam to challenge Chinese open-weight models

Nvidia-backed Reflection AI introduced Beam on October 5, entering the competition for open-weight models aimed at coding, reasoning and agentic work. Founded in 2024 by former DeepMind researchers Misha Laskin and Ioannis Antonoglou, Reflection is positioning the release as an alternative to systems from Chinese developers such as DeepSeek, Qwen and Z.ai, according to Reuters.

Beam is a text-only mixture-of-experts model with 501 billion total parameters and 23 billion active parameters. TechCrunch reports a one-million-token context window and training on 23.8 trillion tokens. Reflection says the model competes with GLM-5.2 on advanced reasoning benchmarks while using substantially less inference compute. Those performance and efficiency claims have not been independently verified.

The commercial stakes extend beyond leaderboard rankings. Reflection is targeting organizations that want customized, locally controlled AI systems. Beam gives those buyers another candidate to evaluate, but active parameter counts alone do not establish real-world operating costs. Independent testing of accuracy, latency and deployment requirements will be more useful than treating vendor benchmark claims as settled comparisons.

4. DeepSeek reportedly nears a roughly $12 billion funding round

DeepSeek is close to securing at least 80 billion yuan, approximately $11.93 billion, in new funding, Reuters reported on October 6, citing Bloomberg News. Tencent and battery maker CATL reportedly committed among the largest amounts. Bloomberg’s sources said investor demand exceeded an initial target of about 50 billion yuan and that the final total could approach 100 billion yuan.

The distinction between reported negotiations and a completed transaction matters: Reuters said it could not immediately verify Bloomberg’s account, and DeepSeek, Tencent and CATL did not immediately respond to requests for comment. The funding should therefore not be treated as closed or its final size as established.

The report follows DeepSeek’s September release of V4.1-Flash and its partnership with Huawei to develop programming tools optimized for Ascend AI chips. If completed at the reported scale, the financing would strengthen a major Chinese competitor as model development increasingly depends on sustained access to capital, computing capacity and a supporting software ecosystem.

5. Power shortages threaten to slow the AI supply chain unevenly

A Morgan Stanley assessment highlights a constraint that model announcements and funding totals cannot solve by themselves: electricity. Reuters reported on October 5 that the bank estimates a 34% net power shortfall for U.S. data-center developers through 2028, equivalent to 32 gigawatts, even after allowing for measures including on-site generation and fuel cells.

The bank does not currently see those bottlenecks threatening its 2027 forecasts for Nvidia or Broadcom, citing deployment visibility, geographic expansion and coordination across the supply chain. It sees greater exposure for memory, optical, power-management and analog-component suppliers if customers postpone deliveries or cancel orders because installed computing capacity cannot be brought online.

These are analyst estimates, not a guaranteed outcome. Nevertheless, the warning separates demand for AI from the ability to deploy it. For businesses planning infrastructure, power availability and commissioning schedules belong alongside chip supply and model performance in any assessment of when new capacity will actually become usable.

The common thread is execution: stronger models and larger investments matter only when organizations can deploy them reliably, identify their outputs and account for what their agents do.

Oct 05

☁️ AI Weather Report — Top 10 Models for Coding Value — October 06, 2026

Welcome to the AI Weather Report for October 06, 2026. This daily report ranks the top 10 AI models for coding by bang for the buck — a combination of raw coding capability and API pricing.

📊 Today’s Top 10 Rankings

#ModelProviderCapabilityCost /M tokensValue Score
🥇 1 mistral-nemo mistralai 62/100 $0.0272 2275.2
🥈 2 l3-lunaris-8b sao10k 58/100 $0.0475 1221.1
🥉 3 gpt-oss-20b openai 78/100 $0.0720 1083.3
4 mistral-small-24b-instruct-2501 mistralai 72/100 $0.0725 993.1
5 llama-3.1-8b-instruct meta-llama 62/100 $0.0725 855.2
6 laguna-xs-2.1 poolside 72/100 $0.1050 685.7
7 gpt-oss-120b openai 93/100 $0.1368 680.1
8 gemma-3-4b-it google 50/100 $0.0875 571.4
9 qwen3.5-9b qwen 72/100 $0.1375 523.6
10 gemma-3-12b-it google 60/100 $0.1250 480.0

📈 Analysis

🏆 Best Value Today: mistral-nemo scores 2275.2 with a capability rating of 62 at $0.0272/M tokens.

What “Value Score” means: Capability score (based on SWE-bench, HumanEval, LiveCodeBench) divided by blended cost per million tokens (25% input + 75% output weights for coding workloads). Free tier models get a massive boost. Higher is better.

📋 All Scored Models (60 total)

#ModelProviderCapabilityCost /M tokValue
1mistral-nemomistralai62$0.02722275.2
2l3-lunaris-8bsao10k58$0.04751221.1
3gpt-oss-20bopenai78$0.07201083.3
4mistral-small-24b-instruct-2501mistralai72$0.0725993.1
5llama-3.1-8b-instructmeta-llama62$0.0725855.2
6laguna-xs-2.1poolside72$0.1050685.7
7gpt-oss-120bopenai93$0.1368680.1
8gemma-3-4b-itgoogle50$0.0875571.4
9qwen3.5-9bqwen72$0.1375523.6
10gemma-3-12b-itgoogle60$0.1250480.0
11mythomax-l2-13bgryphe48$0.1025468.3
12command-r7b-12-2024cohere54$0.1219443.1
13granite-4.0-h-microibm-granite38$0.0882430.6
14ministral-3b-2512mistralai42$0.1000420.0
15nova-micro-v1amazon45$0.1137395.6
16qwen3-32bqwen88$0.2300382.6
17mistral-small-3.2-24b-instructmistralai78$0.2109369.8
18qwen3-coder-30b-a3b-instructqwen84$0.2275369.2
19qwen-2.5-7b-instructqwen60$0.1750342.9
20gemma-4-26b-a4b-itgoogle72$0.2104342.2
21qwen3.5-flash-02-23qwen70$0.2112331.4
22qwen3-30b-a3b-instruct-2507qwen82$0.2500328.0
23gpt-oss-safeguard-20bopenai77$0.2437315.9
24nemotron-3-nano-30b-a3bnvidia50$0.1625307.7
25nova-lite-v1amazon58$0.1950297.4
26gemma-4-31b-itgoogle74$0.2775266.7
27seed-1.6-flashbytedance-seed64$0.2437262.6
28gpt-5-nanoopenai82$0.3125262.4
29step-3.5-flashstepfun60$0.2500240.0
30seed-2.0-minibytedance-seed72$0.3250221.5
31qwen3-235b-a22b-2507qwen96$0.4350220.7
32nemotron-3-super-120b-a12bnvidia76$0.3575212.6
33llama-3.1-70b-instructmeta-llama82$0.4000205.0
34llama-3.3-70b-instructmeta-llama84$0.4300195.3
35llama-3.2-1b-instructmeta-llama30$0.1575190.5
36glm-4.7-flashz-ai60$0.3151190.4
37gemma-3-27b-itgoogle68$0.3575190.2
38gpt-4.1-nanoopenai60$0.3250184.6
39llama-3.2-3b-instructmeta-llama48$0.2600184.6
40gpt-4o-miniopenai74$0.4875151.8
41hy3-previewtencent68$0.4950137.4
42command-r-08-2024cohere60$0.4875123.1
43deepseek-chatdeepseek90$0.8359107.7
44qwen3-next-80b-a3b-instructqwen90$0.8500105.9
45qwen3-coderqwen85$0.8250103.0
46qwen3-next-80b-a3b-thinkingqwen93$0.937599.2
47deepseek-v4-flashdeepseek91$0.967594.1
48qwen-2.5-coder-32b-instructqwen86$0.915094.0
49hermes-3-llama-3.1-405bnousresearch78$1.0078.0
50dolphin-mistral-24b-venice-editioncognitivecomputations52$0.725071.7
51gpt-4.1-miniopenai76$1.3058.5
52deepseek-r1deepseek95$2.0546.3
53gemini-2.5-flashgoogle86$1.9544.1
54nova-pro-v1amazon70$2.6026.9
55gpt-4.1openai90$6.5013.8
56gpt-5openai97$7.8112.4
57gemini-2.5-progoogle94$7.8112.0
58gpt-4oopenai88$8.1310.8
59command-r-plus-08-2024cohere68$8.138.4
60claude-sonnet-4anthropic96$12.008.0

Generated 2026-10-06 02:00 UTC · Data from OpenRouter API and public benchmarks · Bang-for-Buck = Capability / Cost