Nemotron 3.5 Bolt 74B vs DeepSeek V4 Pro: ambos abiertos, uno cabe en 24GB

N​VIDIA's N​emotron 3.5 Bolt 74B-A9B 24GB and DeepSeek V4 Pro are on the same side of the open-versus-closed line and nowhere near each other on the hardware line. Both publish weights you can download, both permit commercial use, and both can be served without asking anyone's permission. One of them is 74 billion parameters with 9 billion active, quantised to fit in 24 gigabytes so it runs on a laptop-class machine next to your other applications. The other is 1.6 trillion parameters with 49 billion active, and the phrase "deployment footprint" does not mean what it means for the first one.

Esa diferencia es más determinante que las puntuaciones de los índices, y las puntuaciones de los índices ya están a 15 puntos de distancia. A continuación, la comparación que un equipo realmente tiene que hacer: no qué modelo es mejor en abstracto, sino cuál de estos dos productos de pesos abiertos puedes operar físicamente, con qué calidad y por cuánto dinero.

Dos formas muy distintas de «open»

Worth stating plainly before any of it, though, because the two open-weight halves did not arrive the same way. D​eepSeek published V4 Pro as a product, with a card and a rate card. N​VIDIA has not announced N​emotron 3.5 Bolt 74B-A9B 24GB at all. The files simply appeared: the 24GB quantised repository on 7 October 2026, its GGUF conversion hours later, a BF16 checkpoint on 9 October, the LiLiCorr speculative drafter on 12 October. Sixteen days of silence on N​VIDIA's own channels is the state of things, so the verifiable half of this comparison is what a repository can prove — licence, parameter counts, file formats, the serving recipe on the card — and the unconfirmed half is everything a launch post would have carried. The throughput and agentic figures below are N​VIDIA's own and are marked as such, because there is no third party except one index to check them against.

Empieza por la licencia y el checkpoint, porque ahí es donde realmente se toma una decisión sobre modelos de peso abierto.

• Licence — N​emotron 3.5 Bolt 74B-A9B 24GB: OpenMDW-1.1, a permissive model-weights licence with an explicit commercial-use carve-in; N​VIDIA's card states the model is ready for commercial or non-commercial use. DeepSeek V4 Pro: MIT, which is about as unrestricted as a weights licence gets.

• What is actually publishable — N​emotron 3.5 Bolt 74B-A9B 24GB: the 24GB IQ2 build, a BF16 checkpoint before quantisation, GGUF builds for llama.cpp, and a separate LiLiCorr speculative drafter. DeepSeek V4 Pro: the full-weight checkpoint and its derivatives.

• Where the checkpoint came from — N​emotron 3.5 Bolt 74B-A9B 24GB: compressed from N​VIDIA's N​emotron 3.5 Super VL 120B-A12B parent using Star Elastic, a router-guided architecture search, then quantised with quantisation-aware distillation. It is a distilled child, not a from-scratch model, and N​VIDIA's own research page argues explicitly that compressing a large model into a fixed budget beats training a small one from scratch. DeepSeek V4 Pro: a natively trained frontier-scale mixture-of-experts.

Esa tercera viñeta es la que decide la mayoría de las evaluaciones que leerás. Cualquier comparación de benchmark entre estas dos es una comparación entre una filosofía de diseño de compresión agresiva y una filosofía de diseño de escala, y las dos no convergen en las mismas cargas de trabajo.

Qué mide cada uno en

Artificial Analysis ha ejecutado ambos, sobre la misma revisión actual de su índice, que es la única razón por la que una frase como esta puede escribirse siquiera.

• Intelligence Index — N​emotron 3.5 Bolt 74B-A9B 24GB 21, fourth of 67 models in its open-weight size class. DeepSeek V4 Pro 36, ninth of 116, measured on the 0813 build in the max configuration.

• Cost per Intelligence Index task — N​emotron 3.5 Bolt 74B-A9B 24GB: not applicable, because the panel records no public API for it and there is no metered price to divide by. DeepSeek V4 Pro: 0.67 dollars measured on the panel's own weighted basis.

• Output speed — N​emotron 3.5 Bolt 74B-A9B 24GB: N​VIDIA reports 83.99 tokens per second on AIPerf SPEED-Bench-Coding with its drafter, on a Surface Laptop Ultra. DeepSeek V4 Pro: 96.3 tokens per second measured by Artificial Analysis against the hosted API. The two are not the same measurement and should not be subtracted from each other; what they establish is that both are fast enough that speed is not your constraint.

• Verbosity — N​emotron 3.5 Bolt 74B-A9B 24GB generated 230 million tokens during the panel's index run, against a median of 100 million for its class — the panel's own note calls it very verbose. DeepSeek V4 Pro generated 160 million against a median of 130 million. On a metered API that difference is a line item; on a local box it is electricity and wall-clock.

The 15-point index gap is real and is not closed by the fact that the smaller model is cheaper to run. What closes it, in the workloads where it closes at all, is the vendor's own benchmark set — and that set is worth reading with the right suspicion. N​VIDIA's card puts the Bolt model at Terminal-Bench 2.1 of 50.98 and BrowseComp of 66.75, ahead of every comparator column printed beside it. Those are N​VIDIA's numbers on N​VIDIA's card, and D​eepSeek is not one of the columns.

El diagrama de despliegue es la decisión real

This is where DeepSeek V4 Pro and N​emotron 3.5 Bolt 74B-A9B 24GB stop being substitutes for one another.

• Room to run. A 1.6-trillion-parameter model at 49 billion active needs a serving cluster. Even heavily quantised, the checkpoint alone is an order of magnitude past what a single workstation can hold, and the active-parameter count means every token touches a substantial slice of memory bandwidth regardless of the quantisation scheme. The Bolt model was designed against a stated ceiling — N​VIDIA's architecture search targets a footprint below 24GB at batch size one and 128K context — and the search is run with the downstream quantisation already accounted for.

• Minimum viable serving stack. N​emotron 3.5 Bolt 74B-A9B 24GB wants a specific vLLM nightly with the v2 model runner enabled, a downloaded reasoning-parser plugin, b12x kernels, and explicit MoE, Mamba and attention backend flags; llama.cpp covers the GGUF path. DeepSeek V4 Pro is served by the usual inference stacks at distributed scale, which is a different kind of project with a different kind of on-call rotation.

• Entradas. El modelo Bolt acepta texto, imágenes y video, y la receta de referencia limita los fotogramas de video y aplica una tasa de poda. La lista de modalidades publicada de DeepSeek V4 Pro es texto de entrada, texto de salida: su escala de mezcla de expertos se dedica al razonamiento, no a la percepción.

• Context. 262,144 tokens for the Bolt model against roughly a million for DeepSeek V4 Pro. If your workload is a repository, a contract set or a long agent transcript, that is a real four-times difference and it is in D​eepSeek's favour.

Put plainly: if you have a datacenter and a workload that needs frontier reasoning, DeepSeek V4 Pro is the better model and the index says so. If you have a laptop, a Spark box or a workstation, and the requirement is that the model runs locally alongside everything else, N​emotron 3.5 Bolt 74B-A9B 24GB is not a compromise on that requirement — it is the only one of the two that meets it.

Lo que el modelo más pequeño te aporta más allá del disco

Tres cosas que no aparecen en una puntuación de índice y sí aparecen en una revisión de despliegue.

• It runs with the network cable out. A 24GB checkpoint on local storage has no availability dependency, no rate limit and no provider incident. For an agent that runs on a machine you own, that is the whole design goal and the reason N​VIDIA compressed a 120B parent instead of shipping something at full precision.

• Speculation is cheap and safe here. The LiLiCorr drafter adds under a gigabyte on disk and N​VIDIA reports mean acceptance lengths of 4.68 greedy and 4.25 probabilistic at seven drafted tokens and concurrency 64. Because the target verifies every draft, a bad guess costs speed rather than accuracy — the output distribution is preserved. That property is what makes shipping a speculative path reasonable on a local box, and it is documented on the model card.

• Todo el stack es inspeccionable. Los pesos, la configuración de cuantización, el modelo borrador, la receta de servicio y el informe técnico son públicos. Cuando el modelo falla, puedes leer la configuración en lugar de abrir un ticket.

Evaluando honestamente el costo de las dos opciones

La comparación de precios no es simétrica, y fingir que lo es sería el error más fácil de cometer en esta página.

• N​emotron 3.5 Bolt 74B-A9B 24GB has no per-token price because it has no metered endpoint. Its cost is hardware you buy or already own, plus power, plus the engineering hours to stand up the serving recipe. Once that is paid, marginal cost per token is close to zero and stays there.

• DeepSeek V4 Pro is metered. On OrcaRouter it is routed at D​eepSeek's published list, 0.66 dollars per million input tokens and 1.98 dollars per million output tokens, with no markup added on top — so a vendor price move is live on our side the same day. For a workload in the millions of tokens a day, that is a budget line, not a rounding error.

• El punto de equilibrio no es una cantidad de tokens, sino un perfil de utilización. Un modelo local que permanece inactivo es puro capex. Un modelo alojado que ejecuta una solicitud a la semana es casi gratis. El punto de cruce se produce cuando los tokens que medirías superan el coste amortizado de la máquina, y la única manera de saber dónde está ese punto para tu tráfico es medir una semana de él.

También hay una tercera opción que casi no cuesta nada probar, y es que OrcaRouter enruta DeepSeek V4 Pro en una sola clave junto con más de 200 otros modelos, con conmutación automática por error entre proveedores. Antes de dedicar una estación de trabajo a una compilación local de 24 GB, comparar la salida de esa compilación con la de un modelo de frontera de pago por uso en tu propio conjunto de prompts cuesta unos pocos dólares y resuelve la cuestión de la calidad empíricamente en lugar de basarse en la solidez de una tabla de índices.

La versión de esta decisión de la que nadie se arrepiente

Run both, in the pattern each is good at. DeepSeek V4 Pro for the requests where reasoning depth or a million-token context is the point, and where the metered cost is visible and acceptable. N​emotron 3.5 Bolt 74B-A9B 24GB for the high-volume, private, multimodal or offline traffic, where the model being local is the requirement rather than a cost optimisation.

La comparación que la gente sigue empeñada en hacer —un número de índice frente a otro— es el marco menos útil aquí, porque los dos productos no se crearon para competir. Uno se creó para ser lo mejor posible sin una restricción de huella. El otro se creó para ser lo mejor posible dentro de 24 gigabytes. Ambos son abiertos, ambos son legítimos, y es exactamente por eso que la decisión es tuya y no de un proveedor que la tome por ti.

Comparaciones como esta son también donde se manifiestan los modos de fallo: los incidentes de IA documentados y los fallos de agentes que hay detrás de los modelos de esta clase se recopilan en the OrcaRouter AI Incident Archive, donde cada entrada incluye sus propias fuentes, su gravedad, su grado de confianza y su indicador de disputa.