Sign in

Head to head · Open-weight models you run yourself · October 2026 research run

DeepSeek-V4-Flash-0731 vs EXAONE 4.5 33B

DeepSeek-V4-Flash-0731 and EXAONE 4.5 33B score within a point of each other on agent readiness, 61.9 (C) and 61 (C). EXAONE 4.5 33B leads on agent interface and documentation and upkeep. Both do open-weight models you run yourself.

Best open-weight models you download and run yourself for AI agents · All 6359 open weights comparisons

DeepSeek-V4-Flash-0731DeepSeek · Open-weight model C61.9/100#33 of 160 open-weight models EXAONE 4.5 33BLG AI Research · Open-weight model C61/100#61 of 160 open-weight models

Which one, for what

DeepSeek-V4-Flash-0731 C

Best for Agents or teams with a multi-GPU server, or a provider account, who want DeepSeek's V4 agent model under MIT with a 1M context

Ahead on

  • Reach, 70 against 65

Watch for

The smallest 4-bit build is 156.4 GB, so it needs about 188 GB and doesn't fit a 128 GB machine

EXAONE 4.5 33B C

Best for Researchers and students working on documents, Korean or multimodal agents who don't need commercial rights

Ahead on

  • Agent interface, 85 against 75
  • Documentation and upkeep, 50 against 38

Watch for

Research and education only. No commercial use of the model or its output

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Documentation and upkeep EXAONE 4.5 33B +12

DeepSeek-V4-Flash-0731 38

Model card 8, the card and the DSpark card it names give architecture, parameters and context, with intended use, limitations, languages and cut-off missing. Training data 5, a general description in the report with no cut-off. Last change 10, a card edit on 1 August 2026, 93 days before 2 November, so past 90 and within 180; deepseek-recipe 0.2.0 (20 October) stays read as a library-wide release, not a change to this model's files. Responsiveness 0, DeepSeek answered none of the 61 discussions, the newest opened on 17 October. Version policy 10, the preview and DSpark copies keep their own repositories and the dated 0731 release says what it supersedes. File health 5, the files agree with each other, but reports of tool-call errors and looping and a pull request for a missing chat template wait without a reply from DeepSeek.

EXAONE 4.5 33B 50

Model card 20, Uses (the quickstart's question answering, document and agent examples), limitations, the architecture, the parameters, the context, the languages (in the metadata) and a knowledge cut-off of December 2024 are all on the card. Training data 5, A knowledge cut-off and one line on excluding personal, harmful and biased content, with no sources or sizes. Last change 10, the main repository last changed on 6 July 2026 (dependencies moved to official releases), 127 days before this check, and the FP8 repository on 9 August, 93 days. Responsiveness 0, none of the 5 most recent discussions that aren't pull requests (13 April to 4 May 2026) has a reply from an LGAI-EXAONE member, counted on 10 November by author flags and dates with a script, text not read. Version policy 5, EXAONE 4.0 and 4.0.1 stay in their own repositories and 4.5 came out in new ones, with no dated note on the older cards that we read. File health 10, Config, template and card agree on the architecture and the 262,144 tokens, and a July commit moved the dependencies to released versions. 20 + 5 + 10 + 0 + 5 + 10 = 50.

Agent interface EXAONE 4.5 33B +10

DeepSeek-V4-Flash-0731 75

Chat template 10, there is no Jinja template and the format is given as Python to run. Tool-calling format 10, DeepSeek documents how tools are declared and returned and ships its own parser, but only vLLM's recipe names a runtime parser, so the 25 line isn't met. Trained tool calling 20, several invoke elements in one call block. Structured output 15, DeepSeek's deepseek-recipe library documents JSON Schema response formats for V4 prompts. Reasoning control 10, chat mode with no thinking and three effort levels in thinking mode, with reasoning returned as reasoning_content. Context 10, 1M tokens trained.

EXAONE 4.5 33B 85

Chat template 20, chat_template.jinja in the repository (5,930 characters). Tool-calling format 25, the card's vLLM command names the parser, hermes with auto tool choice, and its agentic example calls tools from MCP servers through the OpenAI-compatible API, so declaration and return follow that API. Trained tool calling 20, the card reports agentic tool use among its language results and shows an agent run with several tool servers, and the template renders every call in a turn. Structured output 0, Nothing on the card about JSON or schema-constrained output. Reasoning control 10, enable_thinking is on by default and the card says to set it false for the non-reasoning mode, with an example of each. Context 10, 262,144 tokens per the card's configuration table and config.json, and the card's serve commands run at that length. 20 + 25 + 20 + 0 + 10 + 10 = 85.

Reach DeepSeek-V4-Flash-0731 +5

DeepSeek-V4-Flash-0731 70

Memory 5, because the smallest build of 4-bit or better from a runtime's own library is LM Studio's MXFP4 GGUF at 156.4 GB, which with a fifth added needs 187.7 GB, more than our 107.4 GB machine and within 256 GB. DeepSeek's own release, 166.9 GB, sits in the same band. Runtimes 25, the cap, from vLLM v0.32.0, SGLang v0.5.22, llama.cpp release b11771, Transformers 5.20.0 and LM Studio's library, with Ollama not counted since its library entry is retired with no models pushed. Builds 10, DeepSeek ships one build, the FP8 and FP4 release itself. Terms 15, MIT with nothing added. Download 10, not gated. Hosted 5, DeepSeek's API moved to V4.1 Flash, but five inference providers on Hugging Face serve this repository live (Together's mapping has gone since 21 October).

EXAONE 4.5 33B 65

Memory 20, the smallest build at 4 bits is LG's IQ4_XS GGUF, 18.03 GB with the 2.57 GB projector, and 20.60 GB × 1.2 = 24.7 GB, within 32 GB. Without the projector, for text only, it would fit 24 GB. Runtimes 15, Released vLLM 0.33.0 has exaone4_5.py, Transformers 5.22.2 has the exaone4_5 model, and llama.cpp b12242 has the architecture, with LG's own GGUF. Three at 5 each. Builds 15, LG publishes GGUF at five quantised precisions, an AWQ build and an FP8 build. Terms 5, Research and educational use only. Download 10, the Hub API reports gated false for all four repositories on 10 November 2026, so the files download with no account and no click-through, though the licence says downloading is acceptance. Hosted 0, Hugging Face maps no inference provider to it, and the card names no API that serves it. 20 + 15 + 15 + 5 + 10 + 0 = 65.

Score by category

CategoryWeight this runDeepSeek-V4-Flash-0731EXAONE 4.5 33BEdge
Capability35%pendingpendingpendingnot scored in this run
Reach25%257065DeepSeek-V4-Flash-0731 +5
Risk20%205050even
Agent interface10%107585EXAONE 4.5 33B +10
Documentation and upkeep10%103850EXAONE 4.5 33B +12
Negative events≤1500
TotalC 61.9/100C 61/100

Facts side by side

FactDeepSeek-V4-Flash-0731EXAONE 4.5 33B
KindOpen-weight modelOpen-weight model
VendorDeepSeekLG AI Research
Hosted endpointno (local only)no (local only)
Transports
AuthNoneNone
PricingFreeFree
x402nono
LicenceMITEXAONE AI Model License Agreement 1.2 - NC (research and education only)
Read-only variant documentednono
llms.txtnono
Last release2026-07-312026-07-06
Terms last updatedno document linkednot read yet
Privacy policy last updatedno document linkedno document linked
Agent reviews2.5/5 (4)none

How the scores have moved

DeepSeek-V4-Flash-0731

EXAONE 4.5 33B

No score change since it was listed.

Verdicts

DeepSeek-V4-Flash-0731

A long-context agent cousin under MIT with documented tool calls, three reasoning effort levels and support in every major runtime but Ollama. It needs about 190 to 200 GB of memory, though five inference providers serve it. The main risks are prompts built from DeepSeek's Python, no published safety results and no answers in its repository.

EXAONE 4.5 33B

A well-documented vision-language model with a full set of builds, tool calling, switchable reasoning, a 256K context and a stated knowledge cut-off. The licence decides whether you can use it: research and education only, output included, and LG can change the terms. Nothing answered on the repository and no safety results.

Before you call either

DeepSeek-V4-Flash-0731

  1. vllm serve deepseek-ai/DeepSeek-V4-Flash-0731 --trust-remote-code --kv-cache-dtype fp8 --block-size 256 --data-parallel-size 4 --enable-expert-parallel --moe-backend deep_gemm_mega_moe --attention-config '{"use_fp4_indexer_cache": true}' --speculative-config '{"method":"dspark","num_speculative_tokens":7,"draft_sample_method":"greedy"}' --tokenizer-mode deepseek_v4 --tool-call-parser deepseek_v4 --enable-auto-tool-choice --reasoning-parser deepseek_v4 (the card's 4x GB300 example plus the parser flags from vLLM's recipe)
  2. Build prompts with encoding_dsv4.py, deepseek-recipe's V4 encoder or vLLM's --tokenizer-mode deepseek_v4, since the repository has no Jinja template
  3. Tool calls come back in a DSML tool_calls block with invoke elements, string parameters carry string="true" and JSON values string="false"
  4. Set temperature 1.0 and top_p 0.95 for agent work, and allow up to 384K output tokens at reasoning effort high or max
  5. Without the hardware, call it through a Hugging Face inference provider (DeepInfra, Novita, Baseten, Scaleway, Featherless), not DeepSeek's API, where deepseek-v4-flash now means V4.1 Flash
  6. LM Studio's library build is a 156.4 GB MXFP4 GGUF and LM Studio states 156 GB minimum system memory

EXAONE 4.5 33B

  1. Don't use it, or anything it writes, commercially. The licence covers the output too
  2. Serve it with vLLM and --reasoning-parser qwen3 --enable-auto-tool-choice --tool-call-parser hermes
  3. Set enable_thinking false for a direct answer. Thinking is on by default
  4. Use temperature 1.0, top_p 0.95 and presence_penalty 1.5 in general, and temperature 0.6 with top_k 20 for documents and Korean
  5. For llama.cpp, load a projector file from LG's GGUF repository with the weights for image input
  6. Its knowledge stops at December 2024

Questions

Which is better for AI agents, DeepSeek-V4-Flash-0731 or EXAONE 4.5 33B?

DeepSeek-V4-Flash-0731 and EXAONE 4.5 33B score within a point of each other on agent readiness, 61.9 (C) and 61 (C). EXAONE 4.5 33B leads on agent interface and documentation and upkeep.

Other comparisons with DeepSeek-V4-Flash-0731 or EXAONE 4.5 33B

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.

Get an auditSee a sample report